Chapter 6 Epidemiology and Surveillance

6
220
I The Johns Hopkins and the International Federation of Red Cross and Red Crescent Societies
Epidemiology and surveillance
Treguine Refugee Camp, Chad, November 2006 684 children were vaccinated and 1302 women were vaccinated against tetanus.International Federation
Public health guide for emergencies I 221
Description
This chapter provides an overview of the key epidemiological principles and the epidemiological tools needed in managing emergency public health programmes. The main goals are to reduce morbidity and mortality among disaster affected and displaced populations.
Objectives
To provide a basic understanding of key epidemiological principles and terminology; To develop skills for defining and calculating indicators; To describe standard methods for conducting needs assessment; To define the steps for setting up a surveillance system for emergency situations; To describe the main principles and practical methods for conducting a population survey; To identify key steps in investigating disease epidemics; To develop skills in analysing and presenting epidemiological information.
Key competencies
To recognise the main constraints in applying epidemiological methods to emergency situations; To calculate key indicators of the health status of a population; To plan a rapid needs assessment; To set up a surveillance system for emergency situations; To conduct a population survey using the appropriate sampling and analysis methods; To investigate an epidemic and apply findings to an epidemic control programme; To analyse and present epidemiological data in a logical manner for use in programme management.
Basic principles of epidemiology in emergencies

Epidemiology is the study of the causes and distribution of disease in human populations. An epidemiological approach helps planners to focus on the main problems of a community rather than of individual patients and to identify measures for improving the health of the community as a whole. Commonly used terms in epidemiology are defined in the glossary at the end of this chapter. Epidemiology can increase the general understanding about a disease and particularly how it is transmitted even when the cause is unknown. In epidemiology, the assumption is that diseases do not occur randomly, but follow predictable patterns that can be studied and expressed in terms of
what, who, where, when, how, why, and what next.
222
The goal of epidemiology is to identify subgroups of the population who are at a higher risk of disease than usual and who will benefit the most from disease specific interventions. Epidemiological information can be used to develop prevention strategies according to: Time (peaks at a particular season); Place (limited to specific geographic areas); or Person (groups at risk). In emergencies, epidemiology has three elements: Descriptive Epidemiology determines the distribution of a disease among displaced populations. It describes the health problem, its frequency, those affected, where, and when. The events of interest are defined in terms of the time period, the place and the population at risk. Examples: Monitoring the health status of a population to detect cholera cases, such as, by age, sex, location, water source and duration of stay in a dispersed population or camps. Conducting a nutritional survey to determine the prevalence of acute malnutrition among children under five. Analytical epidemiology compares those who are ill with those who are not in order to identify the risk of disease or protective factors (determinant of a disease). It examines how the event (illness, death, malnutrition, injury) is caused (e.g. environmental and behavioural factors) and why it is continuing. Standard mathematical and statistical procedures are used. Example: Investigating an outbreak of an unknown disease in a displaced population settlement. Evaluation epidemiology examines the relevance, effectiveness and impact of different programme activities in relation to the health of the affected populations. Example: Evaluating a malaria control programme for displaced populations.
Role of epidemiology in emergencies

Epidemiology in emergencies goes beyond simply understanding how diseases are contracted and spread. Humanitarian relief programmes can be managed better if decision-making is based on epidemiological findings. Relief workers need training to help them collect more reliable information and use it to improve health care for the displaced population. During the emergency response, applied epidemiology can be used to define the long-term needs of communities. Community capacity to respond to emergencies can be strengthened if members are engaged to learn shoe-leather epidemiology including how to map their villages, conduct reliable needs assessments and identify groups who are at a particularly high risk from disease. Developing such skills can also be highly valuable for communities long after emergencies have been resolved. Objectives of epidemiology in emergencies include: Identifying the priority health problems in the affected community; Determining the extent of disease existing within a community; Identifying the causes of disease and the risk factors; Determining the priority health interventions; Determining the extent of damage and the capacity of the local infrastructure; Monitoring health trends of the community; Evaluating the impact of health programmes.
Constraints of epidemiology in emergencies

Constraints in using epidemiology in emergencies include: Poor understanding of basic epidemiological principles and measurement techniques; Rapid turnover of skilled staff; Limited access to a significant fraction of the affected population due to chaos or insecurity; Limited resources for processing information; Difficulty in estimating the population size; Survey samples might not represent all the affected population; Difficulty in investigating long-term needs of an affected population; Lack of support to undertake assessments and address needs of Internally Displaced Persons (IDPs). Note: governments might be part of a conflict and are thus less likely to care for the IDPs while the original mandate of UNHCR does not cover IDPs.
Key epidemiological indicators

Indicators are measures that reflect the state of a population in terms of health, socioeconomic status etc. They might also reflect the process and outcome of existing services. In humanitarian emergencies, indicators are useful for measuring and describing the effects of a disaster on a population and for providing baseline measurements. Later, these measurements will help determine the outcome of the relief response. Indicators may be defined from surveys or an existing health information system. They might be quantitative or qualitative in nature. Quantitative indicators are easily calculated from numeric information such as the total number of people, the number of people according to age and sex etc. Examples of quantitative indicators include: Incidencethe number of cases or events that occur within a defined population that is divided by the total population in which the cases or events occurred in a specific period e.g. the incidence of measles among children. Prevalencethe proportion of the population with a particular condition that is divided by the total number of persons at risk for the condition during that period e.g. the prevalence of HIV/AIDS in a population. Morbidity rateall persons in the population under consideration e.g. belonging to a specific gender or age-group who become clinically ill during the specific period. Mortality ratethe number of deaths occurring in a population in a specific period (usually a year) divided by the number of persons at risk of dying during that period e.g., mortality rate of infants during their first year of life.
Epidemiology has many uses in emergency situations, including: Rapid needs assessment; Demographic studies determining the population size and structure of affected communities in camp settings or dispersed within a host population; Population surveys for determining health status (death rates, incidence/prevalence of disease, nutrition and immunisation status) and assessing programme coverage; Investigating a disease outbreak; Public health surveillance and management information system; Programme monitoring and evaluation.
224
Qualitative indicators that measure peoples attitudes and knowledge are more difficult to measure. These indicators might be critical in explaining unexpected values of quantitative indicators. Social processes influencing health outcomes might also be elucidated by using qualitative indicators. Examples of qualitative indicators include: Awareness of the value of immunisationlow awareness may explain the high incidence of measles in a population living within five kilometres from a health facility; Adherence to preventive interventions against HIV/AIDSpoor compliance from youths in preventive interventions (e.g.; A lack of understanding about the Abstinence, Be Faithful, Use a Condom (ABC) programme)) might explain the increasing prevalence of HIV/AIDS in a population; Equity in distribution of resourcesinequitable distribution of food might explain the increased mortality detected in a subgroup of a population; Barriers to seeking treatment for malariabarriers to seeking treatment such as unaffordable health services, might explain an increase in malaria-specific mortality. The following qualitative indicators are commonly used for assessing programme outcomes: Access: the proportion of the target population that can use the service or facility; Coverage: the proportion of the target population that has received service; Quality of services: the actual services received compared with the standards and guidelines; Availability: amount of services compared with total target population. This should be based on minimum standard requirements. The following table summarises the quantitative and qualitative indicators that can be used to evaluate an emergency health programmes process and outcome.
Table 6-1: Quantitative and qualitative indicators for emergency health programmes
Indicator Examples
Health policy (might be difficult to measure)
Demographic profile
Health status
Programme inputs
Degree of political commitment Adherence to agreed/national case definitions and treatment protocols Level of community participation Degree of inter-sector collaboration Equity in distribution of resources Inter-agency co-ordination Adherence to universal precautions against HIV/AIDS Adherence to minimum standards of the Sphere Project Estimated size and structure of displaced population: Age and sex composition Migration patterns (proportion moving in and out) Proportion of high-risk groups Ratio to local resident population Rate and causes of death (crude, infant, under fives, maternal) Incidence and prevalence of common disease Rate of under-five malnutrition Availability of the following resources: Facilities and equipment (health centre, beds) Staff (beneficiary population, local, expatriate) Basic supplies (food, shelter material, domestic equipment) Energy sources (fuel, charcoal) Transport
Indicator
Examples
Programme process
Access, coverage, and quality of the following services: General food distribution and supplementary feeding Potable water supply Latrine construction Immunisation Reproductive Health Health services at all levels i.e. community health, peripheral clinics/second and tertiary facilities
Estimating population size

A reliable estimate of the total population size including age and sex distribution is important for the following reasons: To be aware of the true population that is at risk of death and disease including the host population; To have a census for planning and political reasons (required by host authorities, donors, media, etc.); To estimate the amount of basic needs required (food, water, shelter, health resources); To draw budgets for relief programmes and estimate needs in recovery phase; To calculate the value of indicators for programme monitoring and evaluation; To plan long-term solutions including contingencies and future mitigation.
Different methods for estimation

Estimates of the total population, the population of children aged less than five years or women of child-bearing age etc. represent critical denominators that are required for calculating indicators. The most critical denominator is the estimate of the total population. This estimate must be valid since it provides the basis for all planning in humanitarian programmes. These estimates might exist from a prior registration exercise or census. However, they might be grossly unreliable e.g. over-stated or understated especially during disasters with major displacements. One might either accept the best available data or apply current epidemiological knowledge about the effects of similar disaster situations on the population structure. The ideal method for estimating the population size is by a census or a registration system, which can only be carried out several weeks or months after the relief operation has been established.
Acute emergency situations

This can be most challenging since many decisions have to be made on incomplete, inconclusive and inaccurate information and data. It requires very experienced professionals to lead the operations. Epidemiological procedures can be used to get better estimates of the population size and structure. For better estimation of the population size during the acute emergency phase, the following surveys or sampling methods can be used for the rapid assessment: Water Usage: Determine the total amount of water the whole population in the dispersed population/camps consumes in one day. Then, interview a sample of people at their household or water collecting point to estimate the average amount of water used by each individual: If 200,000 litres of water are consumed in one day and individual water usage is
226
estimated as 20 L/person/day, the total population in the dispersed population/camps should be 200,000/20 = 10,000 people. Note: The total food distributed and individual food baskets may be used to estimate the total population in the same way as water usage. However, these estimates should be interpreted carefully since food rations may be collected for sale or families may collect more than one ration. Nutrition Screening: Screen and count a specified fraction of children under five years. In emergencies, the number of women, children and elderly is very high. If birth records are not available, use the cut-off height for all children aged less than five years as 110 cm. Assuming that the under-five children make up about 20% of the total population, multiply the estimated fraction of under-fives by five to estimate the fraction of the total population for the total population estimate. Note: Mass immunisation campaigns can be used to estimate the population size in a similar manner assuming that the immunisation coverage is 90% or more. Aerial Mapping: Maps are useful tools for gathering additional information. A map can be used for sampling people from various ethnic and socio-economic groups for interviews or for sampling households for rapid surveys and for planning and evaluating programmes. If no maps exist, one can either take a photograph of the settlements while flying or manually sketch maps to locate the affected population. The population might either be settled in its own camps or integrated within the host population. Begin with a tour around the boundary of the location(s) to define its approximate shape and the maximum and minimum length and width. The key landmarks (e.g. river, lake), the roads and any Primary Health Care facilities around the catchment area should be included in the map (see Figure 6-1 below). If possible, the varying population density within the location(s) should be shown. Make a rough estimate of the population size using this information or continue to step 4. Note: Aerial mapping is indeed useful, but in floods, where many landmarks are hard to find, A Global Positioning System (GPS) is convenient. However, without appropriate scaled maps and their coordinates, it will not work. Such maps and pilot charts are used by military and civil aviation.
Figure 6-1: Map of catchment area
Note: The legends (symbols and colours representing structures and boundaries) should be consistent and recognisable for all maps. Maps of sub-catchment areas might also be drawn to show varying target groups for different primary health care services.
Determining the populations size and composition: Divide the entire map into sections containing approximately equal numbers of households. To estimate the number of households in the entire location, count the number of households (shelters or cooking fires) in a typical section and multiply this by the total number of sections. Carry out convenience sampling and select a reasonable number of households (e.g.,
Note: The above-mentioned convenience (or non-probability) sampling is useful for making crude estimates of the population size/composition and possibly for identifying the immediate health needs during the rapid assessment. Results from convenience sampling are biased and not representative of the entire population. They cannot be used for comparison with results from other surveys. Where possible, probability sampling surveys should be organised as soon as possible to obtain more reliable results. See the section on population surveys for details about probability sampling methods.
Post-emergency situations
To estimate population size during post-emergency situations, other techniques can be used if the information from census or registration exercises is unreliable: Participatory mapping of the catchment area can be done by inviting a group of the affected population to sketch a map of the entire community on the ground or on large paper. They should first be asked to define the physical boundaries of the location of the affected community (see Figure 6-1 above) and the location of all key landmarks (e.g. rivers or lakes, roads, health facilities/services, water pumps, cemeteries etc). Distances should be shown as accurately as possible. They should be asked to identify where different ethnic communities and the most vulnerable group(s) (e.g. the poorest or most malnourished) are located in the map of the catchment area. Note: The above approach can easily capture the population in camp settings. However, when the affected community is dispersed within a host population, it is important to invite members from the disaster affected as well as the host population to either separately or jointly participate in the mapping exercise depending on the level of hostility between the two groups. Household registration: If the information from the census or registration exercises is unreliable or more information is required as dispersed population/camps services are set up, household and dispersed population/camps registers should be developed. Reviewing existing administrative records or interviewing key persons can help in designing the registers and in determining the target groups for emergency health services. Community health workers can be trained to visit all households and gather the required information e.g. record the households on the map and register each household members personal details (name, age, sex) and any existing risk factor (malnutrition, illness and disability). Household registers can later be used by health workers to locate vulnerable individuals who are most at risk of disease or death and to target them for specific primary health care interventions. Dispersed population/camps registers can be developed from summaries of household registers. They are useful for identifying the priority health needs of the population in the dispersed population/camps. Examples of dispersed population/camps and household registers are shown in the following table.
International Federation refugee camp Photo: International Federation
fifty) that can be easily reached. Record the number of persons living in each household including their age and sex breakdown. Calculate the average number of persons per household and multiply this by the total number of households. The agesex pyramid can be plotted to show the estimated population structure (see data analysis section for an example an of age-sex pyramid).
228
Table 6-2: Camp register
Camp
Population
# Households
Camp leader
CHW volunteer
Health problems Priority Addressed
Access to health facility
Available transport
Remarks
Omega
1000
150
Jacob
Sarah
sanitation Immunisation Immunisation Malnutrition
no yes
10 km
Donkey carts
Water source not reliable Easily floods in rainy season
Delta
1075
210
Noah
Adam
yes no
5 km
none
Table 6-3: Community health worker household register

Registration date: Prior occupation: Camp/zone: Health problem Date of death 23/7/68 Teacher Omega/blue
Household ID: 02/90/12 Head of household: Aladdin Ali
Relation to HoH
Name
DOB/Age
Sex
Migration
Risk factor
Remarks
wife
Delilah
17
pregnant
local beliefs
son father
Sinbad Ali Baba
3 months 58
M M
diarrhoea TB 20/6/68
not breastfeeding
Poor access to formal health care, low literacy
Caution: Keely et al. (2001) cite several potential reasons for the overestimation of the number of beneficiaries.19 These include hiding people who are not Internally Displaces Persons (IDPs) or legitimate refugees, registering more than once to increase food rations, under-reporting of out-migrations and deaths and when members of the local population can attempt to register to access services provided to IDPs or refugees. On the other hand, under-estimations might result if IDPs or refugees who settle outside camps in the local population are hard to find and count, or if persons who are sick or malnourished do not access services that are being used to count beneficiaries.
Calculating rates, ratios and proportions

The most readily available data is usually in the form of absolute numbers (e.g. the total number of measles cases, the total number of diarrhoea cases). Absolute numbers can be used to report on the health of a specific population in a confined area over a short period. However, they cannot be used to compare events within the same population between population groups of different sizes or at different locations because absolute numbers can lead to invalid conclusions. For example, no conclusion can be drawn from reports of twenty-one deaths in Refugee Camp A and fifteen deaths in Refugee Camp B. To understand the significance of these reports and compare the death toll in the two camps, the frequency of deaths must be expressed as fractions such as rates, ratios, or proportions (percentages). These fractions contain a numerator and the best estimated population size as the denominator.
Rate = x/y x factor
The Crude Mortality Rate is the total number of deaths per day and the number of under-five deaths. As with determining population data, counting the number of deaths in an emergency is difficult. It is even harder to count events among beneficiary populations that are not living in camps. Deaths often do not happen at or get reported to service sites such as health facilities or feeding centres. As a result, two additional approaches that can be used to estimate the number of deaths include retrospective mortality surveys and community-based surveillance. Standardised methods for retrospective mortality surveys (such as SMART) are being developed.11 Community-based methods include having volunteers and making home visits to ask about recent deaths, hiring grave watchers to count the number of burials or counting the number of shrouds or other burial material distributed.11 Note: Checchi and Roberts provide an easy to read description of both retrospective mortality surveys and community-based mortality surveillance in Interpreting and using mortality data in humanitarian emergencies: a primer for non-epidemiologists. This paper can be downloaded from the Humanitarian Practice Network. It provides information on sample sizes, sampling methods, question procedures and interpretation of retrospective mortality surveys. These surveys provide relatively good quality information about past mortality experience. However, they should be led only by very experienced people. For prospective, community-based surveillance, Checchi and Roberts recommend a system whereby trained home visitors go to homes within assigned sectors daily or weekly. During their visits, the home visitors inquire about deaths (day of death, age, gender). Home visitors can also use this opportunity to help update information about the size of the beneficiary population (total size, under-five count and eventually age and gender distribution). This system requires substantial supervision and a significant number of home visitors, however. The authors recommend at least one home visitor for every 1,000 beneficiaries. Checchi and Roberts site potential problems with an alternative community-based method: grave-watchers hired to watch burials sites twenty-four hours a day. The potential problems they cite include the difficulty distinguishing new from old graves especially after rain, incomplete knowledge of all burial sites (e.g. some newborn deaths might be buried within or near shelters) and difficulty distinguishing graves of members of the local population from those of the refugee population.21 The notes below show how to calculate these rates assuming there is an existing surveillance system that documents the number of deaths occurring after identifying the total number of the population at risk of death. Both numbers are needed to make the calculation. Note: Retrospective surveys are a good source of information for the number of deaths and the population at risk of death. Experienced persons are needed, however, to calculate the precision of the rates derived from surveys.
This formula expresses the likelihood that a particular event, case, or episode (x) will occur in a specified period of time among a population at risk (y). Examples of rates include the crude mortality rate, cause-specific mortality rate and incidence rate. These are described in greater detail below.
230
Crude Mortality Rate (CMR): The rate of death in the entire beneficiary population (includes both sexes and all ages).1 The most commonly used expression of CMR is the number of deaths per day per 10,000 persons in the population. The population size most commonly used in the calculation is the estimated population size at the mid-point of the time period---the time period that the number of deaths represent (say the number of deaths during a month or a week or a day). The formula for this calculation is provided below: Total number of deaths during the time period Total population at midpoint of time period that is at risk of death
10,000 # of days in time period
# deaths/10,000 persons/day
Under-5 mortality rate (U5MR): The rate of death among children below 5 years of age in the beneficiary population.1 This is a type of age-specific mortality rate. As with the CMR, this rate is most commonly presented as the number of deaths (among < 5s) per 10,000 persons per day. The formula for this calculation is provided below: Total number of deaths in children < 5 years of age during the time period Total number of children less than five years in the population at mid-point of time period that is at risk of death X
10,000 # of days in time period =
# deaths/10,000 persons/day
Crude Mortality Rate and the Under-Five Mortality Rate (U5MR) are typically calculated on a population level; that is, the total number of deaths in the population is counted (surveillance) or estimated (surveys). In contrast to CMR and U5MR, the numerators for other key rates of mortality and morbidity are often based on information collected only at service sites such as health facilities, cholera treatment centres and feeding centres. This is because information about the trends in rates (as opposed to population-based rates) is often sufficient for disease control; obtaining rates on a population-basis for specific causes of death and disease is costly to do on a regular basis. The following rates described below are considered key rates for health information systems to collect and utilise. Cause-specific mortality rates: This rate looks at the number of deaths due to a specific disease or other cause during a specified time period. The rate is reported as the number of deaths due to a specific cause per 1,000 individuals in the population during the time period (weekly during emergency or monthly in the post-emergency phase). Daily reporting should be considered during an epidemic.17 Note: Calculate rates for the causes of death requested by the lead health authority and include on standard mortality reporting forms. The Sphere handbook includes sample weekly mortality surveillance reporting forms that can be used until a standard is defined by the lead health authority.1 Any enquiry on specific deaths needs to include diseases of epidemic potential such as acute watery diarrhoea, dysentery, cholera, measles, malaria and meningitis. Other causes of death typically requested in many emergencies include pneumonia, injury as well as maternal and neonatal illnesses that kill scores worldwide (between 7 to 10 million per year depending on sources). This information is used to select, monitor and evaluate health interventions.
Example: The significance of twenty-one deaths in Camp A and eighteen deaths in Camp B depends on the period that they occurred and the size of the population at risk. Assuming they all occurred over a seven-day period, the crude death rate can be calculated for each camp based on the estimated total populationA (50,000 people) and B (5,000 people) as follows: Crude mortality rate (CMR) Number of deaths x factor Total mid-interval population x time period 21 x 10,000 = 50,000 x 7 15 x 10,000 = 5,000 x 7 0.6 deaths/10,000/day = (indicates a stable situation) 4.3 deaths per 10,000 per = day (indicates a critical situation)
Crude mortality rate for population A
Crude mortality rate for Population B
Note: A factor of 100, 1000, or 10,000 can be used to convert calculated rates into whole numbers. During the initial phase of the emergency, a factor of 10,000 is used for calculating daily death rates in order to detect sudden changes. A crude death rate of one death/10,000/day indicates an acute emergency phase. The post emergency phase begins once the Crude Mortality Rate (CMR) drops below one death/10,000/day. Thereafter, death rates may be analysed once a month using a factor of 1,000 to calculate monthly death rates. To convert CMR expressed as deaths/10,000/day into deaths/1000/month, divide the daily CMR by ten and then by the total days in a month. From the above example: 4.3 deaths/10,000/day = 4.3 10 x 30 = 0.014 deaths/1,000/month
Incidence rates for common causes of mortality and morbidity is the number of new occurrences of an event included on standard morbidity reporting forms (disease, injury, malnutrition, scabies etc.) during a time period in a population at risk for the event. Incidence rates: The Sphere handbook includes sample weekly morbidity surveillance reporting forms that can be used until case reporting is standardised by the lead health authority.1 The specified events may include diseases of epidemic potential such as acute watery diarrhoea, dysentery, cholera, measles, malaria and meningitis. Other events typically requested include pneumonia, injury, skin diseases, sexually transmitted infections including AIDS, and eye diseases. This information is used to select, monitor and evaluate health interventions. The weekly reporting formula for this calculation is provided below: Total number of event during the week Total population at mid-point of the week that is at risk of having the event
*
1,000
= # events/1,000 persons/week*
Meningococcal meningitis is usually calculated per 100,000 persons/week
232
Case-Fatality Rates (CFR) is the proportion of persons with a disease in a specified period that dies from the disease. It is reported as a percent such as the case-fatality rate of persons with cholera in the last week was 35%.
CFR: If cases on a disease arrive at a health facility shortly after the onset of the disease, this rate is a reflection of the quality of care provided in health facilities. Otherwise, this rate is a reflection of the effectiveness of outbreak control measures. The maximum acceptable CFRs are as follows: cholera (1%); Shihe formulgella dysentery (1%); typhoid (1%); meningococcal meningitis (varies; up to 20% during outbreaks).1 Ta used to calculate the weekly case fatality rate is provided below. Total number of people dying from the disease during the last week Total number of people who had the disease during the last week
100
= X%
Health facility utilisation rate looks at the number of out-patient visits per person. The rate is converted to an annual rate even if the time period during which the information is collected on visits is less (usually one week). Health facility utilisation rate: Among displaced populations, an average of 4.0 new consultations/per person/per year may be expected. If the rate is lower than expected, it may indicate inadequate access to health facilities. If the rate is higher, it may suggest over-utilisation due to a specific public health problem (e.g., infectious disease outbreak), or under-estimation of the target population (pg. 268).1 While it is recommended that new visits be counted separately from old visits, it is often difficult to do; during an emergency the total number of visits is typically counted. The formula used to calculate this rate on a weekly basis is provided below. Total number of visit in one week X 52 weeks = # visits/persons/year Total population at mid-point of the week Consultation rates per clinician per day: This looks at the total number of patients seen by each clinician per day on average.1 No effort is made here to distinguish new from repeat consultations.
Total number of patient in one week Number of FTE clinicians in health facility
* *
Number of days health facility open per week
# consultations/ clinician/day
FTE = Full-Time Equivalent. This number represents full-time staffing. Consider a clinic that is open six days per week. If two persons each work all day three days per week, this is one full-time equivalent. If three persons each work all day six days per week, and one person works day six days per week, this is 3.5 full time equivalents. If one person works day six days per week, this is 0.5 full-time equivalents.
Ratio = x/y This formula expresses a relationship between a numerator (x) and a denominator (y), where x need not be part of y. Example: If the estimated size of the displaced population is 20,000 with 8,000 males and 12,000 females: Then, the ratio of males to females Total number of males Total number of females = 8,000 12,000 = 2:3
This ratio is better interpreted by dividing each side of the equation by the value on the left side i.e.: Male: Female = 2/2:3/2 = 1:1.5 Proportion = x/y This formula expresses the relationship between a numerator (x) and a denominator (y), where x is part of y. Proportionate mortality is not a rate by definition. Proportionate mortality looks at the proportion of all deaths in a population during the period due to a certain cause (the denominator is the same as for the Crude Mortality Rate). For example, during one week period, we might find that cholera was responsible for 70% of all deaths in that week. This information is used to select, monitor and evaluate health interventions.
Number of consultations per clinician per day: The Sphere standard is that clinicians are not required to regularly consult more than fifty out-patients per day. The decision to hire additional clinical staff should be triggered if this threshold is regularly exceeded. Clinical staff is defined as a formally trained clinical provider, such as a physician, nurse, clinical officer or medical assistant (page 267).1 Note that Primary Health Care workers (non-physicians) can often see more patients than physicians because physicians tend to manage more complicated cases. The formula used to calculate this rate on a weekly basis is provided below.
234
Example 1: If the 21 deaths in camp A were of patients diagnosed with malaria during the same month: Proportional mortality (%)from malaria in hospitals = Number of deaths due to a certain disease x 100 Total deaths during that period x 100 = 21 x 100 100 = 21%
Example 2: Coverage is also calculated as a proportion as follows: No. of beneficiaries of a service = x 100 Total target population Immunisation (%) Coverage = (No. of children aged 12-23 months who got immunised) (Total number of children aged 12-23 months)
Interpretation of epidemiological data

The main uses of the Crude Mortality Rate and Under-Five Mortality Rate (U5MR) are to (1) Decide if humanitarian assistance is needed; (2) Decide if humanitarian assistance is adequate or should change; and (3) Advocacy. Need for humanitarian assistance: Mortality rates that are double the expected baseline levels are an indicator of significant excess mortality and a justification for initiation or continuation of humanitarian assistance. We recommend the use of Sphere standard thresholds for initiation/continuation of humanitarian assistance as these reflect the doubling of expected baseline levels. The Sphere thresholds are provided in Table below. When baseline rates are unknown, a Crude Mortality Rate (CMR) of >1.0/10,000/day or U5MR of >2.0/10,000/day are used.
Table 6.4. Mortality thresholds used to define emergency situations11.
Agencies Assumed baseline Emergency thresholds
Centres for Disease Control, Mdecins Sans Frontirs Epicentre, Academia UNHCR
Fixed at: CMR: 0.5 per 10,000 per day U5MR: 1 per 10,000 per day Fixed at: CMR: 0.5 per 10,000 per day U5MR: 1 per 10,000 per day
Emergency if: CMR 1 per 10,000 per day U5MR 2 per 10,000 per day CMR > 1 per 10,000 per day: very serious CMR > 2 per 10,000 per day: out of control CMR > 5 per 10,000 per day: major catastrophe (double for U5MR thresholds) Emergency if CMR (U5MR): Sub-Saharan Africa: 0.9 (2.3) Latin America: 0.3 (0.4) South Asia: 0.5 (1.2) Eastern Europe, Former Soviet Union: 0.6 (0.4) Unknown baseline: 1.0 (2.0)
Sphere project
Context-specific CMR (U5MR): Sub-Saharan Africa: 0.44 (1.14) Latin America: 0.16 (0.19) South Asia: 0.25 (0.59) Eastern Europe, Former Soviet Union: 0.30 (0.20) Unknown baseline: 0.5 (1.0)
However, before disseminating or using any assessment data, its quality should be examined for potential biases. Quality of mortality data can vary depending on whether surveillance or surveys are the source. Because health needs assessments are done rapidly, the analysis or conclusions might not be accurate. Triangulation might be performed to improve the quality of findings as described in the table below.
Triangulation: Confidence in the data collected can be improved by cross-checking in various ways: (1) by different sources, (2) about different persons and (3) using different methods.7 By including persons with diverse perspectives and backgrounds on the assessment team, the potential bias that the information exclusively represents a limited point of view (e.g., views of one technical area or gender, outsider perspective only, etc.) can be overcome. By collecting data about and from persons with diverse characteristics (gender, leadership role, service provider, beneficiaries, ethnicity etc.) the bias that the information only represents the interests of a narrow group of people can be overcome. By using several data collection methods (review of secondary information, interviews with individuals, group discussions, observations) the biases that come with particular methods can be overcome. For example, it might be expected that people might change their behaviour during observation; if we only use observation we may never learn about what people do when not observed. If we only use group discussions, important but sensitive information might not be shared. If we rely only on interviews, we may miss important but complex information that can be picked up by observation.
There are specific guides addressing issues of bias, such as the one by Checchi and Roberts discussed previously in this chapter, which is available for download from the Humanitarian Practice Network.21
Judging the quality of mortality surveillance data should only be used by persons with significant experience and expertise should lead or conduct mortality and those who are able to consider the following when interpreting survey data: The Crude Mortality Rate (CMR) and Under-Five Mortality Rate (U5MR) calculations are very sensitive to the number used to estimate the population size (denominator). When interpreting the data, judge whether or not the population size is likely to be overor under-estimated. Analyze how this might affect the true rates. With small populations, CMR, U5MR and other indicators are less reliable and jump within days. To judge the situation requires experience and a great deal of common sense. Some deaths can be missed by any surveillance system especially if death reports are limited to what is collected at service sites. Similarly, a single death can be recorded twice if several sources of data are used to count deaths. When interpreting the data, consider the sources of the death reports (e.g. facility registers only, community-based surveillance, or both) and judge whether or not the number of deaths is likely to be an over- or under-estimate and how this might affect the true rates.
Need to change humanitarian interventions: Sites with ongoing CMR or U5MR at emergency level should consider re-prioritizing the health interventions provided (see Standard 1 for Health Systems and Infrastructure). Use available information about causes and age of deaths to decide how current health interventions need to change, or if other interventions such as food distribution, supplementary feeding, water and sanitation need improving. Advocacy: When the needs of populations in crisis are being ignored by media and political leaders, mortality studies can be used to raise awareness and apply pressure for adequate humanitarian assistance.11
236
Rapid health assessment
Objectives of rapid health needs assessment

Key decisions in a health emergency, such as the need for intervention and the type and size of the intervention, must be made rapidly. Unfortunately, precise and reliable data often require weeks to organise, collect and analyse. The methods used are more complex. This means that some decisions must be made using less precise and less reliable data. The assessment should be started within a short time frame, ideally within the first three days after the event. A rapid needs assessment is carried out to determine:
The magnitude of the emergency and population size affected;

The vulnerable population groups with high risk of death or disease; What are the present health priorities and potential public health problems; Availability of food, water and shelter, clinics and hospitals; Environmental conditions; What the government or military is doing, what the plans are; Presence, plans and activities of international or local organisations; Whether or not external support or intervention is needed; and Critical upcoming hazards, like winter conditions.
Preparing for rapid health needs assessment

Adequate preparations are needed before a rapid assessment can be made. This includes the following steps: 1. Before the field assessment, collect background information about the emergency situations geographic location, the population affected and any political factors. Also collect the pre-emergency health data and information on the existing health system. Try to confirm all information with the UN, the host government and other NGOs; 2. Coordinate with other organisations, government and military and assessments; 3. If a multidisciplinary assessment team cannot be recruited locally, get the proper authorisation (work permits, travel permits, vaccinations etc.) for additional personnel from outside the country. Local representation is absolutely essential and teams made up of expatriates must be avoided; 4. The assessment team should then plan the field assessment as follows: Define the terms of reference and the objectives of the assessment; Based on the nature of the emergency, determine the priorities to be considered; Select how and in what order the information will be gathered. If existing assessment checklists are to be used, coordinate with other agencies. They must be carefully reviewed and adapted to the local situation. All should use the same checklists; Design or adapt the forms for recording and analysing the information collected. All should agree on how and when the information will be reported; Estimate the time frame and the resources needed (stationery, data processing tools, personnel) for each stage of the assessment, such as training field staff and volunteers, collecting and analysing data; Assign specific tasks and responsibilities to each member of the assessment team; 5. Inform all departments within the organisation that need to be directly involved with the assessmentlogistics, finance, human resources etc. Identify the person at headquarters who can be contacted from the field during the assessment;
7. Check the security situation in the field and make contact with local authorities and other organisations; 8. Ensure there is someone based locally to arrange the assessment teams transportation, communication, accommodations and meals; 9. Be aware of the common mistakes that can occur during any assessment. Try to prevent errors by using the actions shown in the following table.
Table 6-5: Preventing common errors during a rapid health needs assessment
Common errors Preventive action
The assessment is poorly coordinated between various NonGovernmental Organisations (NGOs) and excludes the host government and the affected community. The assessment team lacks the expertise needed.
Appoint a team leader to co-ordinate the assessment with the host government, the affected communitys leaders, and other agencies so that the results are shared, are not duplicated and ensure future support of relief activities. Select members of the team with disaster-specific (prior experience), site-specific (geography, language, culture) or speciality-specific skills (epidemiologists, physicians, public health nurses, logisticians, environmental engineers). Psychological trauma and reproductive health are often not assessed. Health related sectors like shelter, water and sanitation, relief needed.
The needs assessment is conducted too late. Collecting information requires time, yet often time is limited. The data collected is often incomplete (due to poor access) or inappropriate (does not cover all the important areas).
Strengthen disaster preparedness by establishing an early warning system for detecting humanitarian emergencies. Plan the field assessment: define the objectives, the relevant information needed and methods for collecting data. Discuss plans with local authorities, community representatives and other agencies. Ensure that one of the main objectives for carrying out the needs assessment is setting up an information system. Collect only data that can be processed. Make better estimates by mapping the location and dividing it into sections. Determine the average family size in selected households of some sections and apply findings to the entire map. Follow epidemiological procedures when carrying out population-based surveys. Involve representatives from the affected population at every stage of the assessment, including drawing conclusions from the local response and outstanding needs.
The data collected is not linked to an ongoing information system. More data is collected than is needed or used. The estimated size of the target populationthe critical denominatoris unreliable. The survey sample does not accurately represent the affected population. The assessment report does not consider the affected populations perceived needs.
6. Collect essential equipment. All members must have double communication means, survival gear and GPS as deemed important, apart from maps, first aid kits etc.;
238
Common errors
Causes of death are incorrectly attributed to the disaster even for slow-onset disasters, such as drought and famine. Assessment team members do not complete their tasks as they are drawn into establishing initial activities. Thus, there is insufficient time for accurate assessments, the assessment period is extended and serious delays in vital action can occur. Information is not shared with government, other agencies and clusters.
Preventive action
Collect background information: interview former staff, local authorities, and the media; review field reports, country profiles, and Internet/Medline. Arrange for a local emergency response team (health, fire, police) to take care of the injured and limit harm from hazards (fire, disease epidemics, etc.) so that the assessment proceeds smoothly.
Discuss before assessment when and how to share common planning and priority settings.
Defining the information needed

Defining in advance what information is needed about the emergency can improve the coverage and quality of the needs assessment. It can also help to identify the sources of information to be contacted on arrival at the site. Many checklists for rapid assessments have been developed, some of which classify indicators under different data categories. They are supposed to guide assessment teams in thinking about the information they might want to collect in the rapid needs assessment as well as to ensure that they have covered the key issues. Assessment checklists should not be considered as simple exercises for filling out data forms. Checklists that are used should first be adapted to the context and culture of the specific emergency. Not all the information in the checklist might be needed or be relevant for every assessment.
Table 6-6: Sample checklist for rapid health needs assessment13
Preparation Obtain available information on the disaster-affected population and resources from host country and international sources. Obtain available maps and aerial photographs. Obtain demographic and health data from host country and international sources. Security and access Determine the existence of ongoing natural or human-generated hazards. Determine the overall security situation, including the presence of armed forces or militias. Determine how much access humanitarian agencies have to the affected population. Demographics and social structure Determine the total disaster-affected population and proportion of children under five years old. Determine age and sex breakdown of the population. Identify groups at increased risk, e.g. children, women, older people, disabled persons, people living with HIV, members of certain ethnic or social groups. Determine the average household size and estimates of female- and child-headed households. Determine the existing social structure, including positions of authority/influence and the role of women.
Logistics Assess transport, fuel and storage of food, vaccines and other supplies, communication. Mortality rates Calculate the Crude Mortality Rate (CMR). Calculate the Under-Five Mortality Rate (U5MR) which is the age-specific mortality rate for children under five. Calculate cause-specific mortality rates. Morbidity rates Determine incidence rates of major diseases that have public health importance. Determine age- and sex-specific incidence rates of major diseases where possible. Available resources Determine the capacity of and the response by the Ministry of Health of the country or countries affected by the disaster. Determine the status of national health facilities including total number, classification and levels of care provided, physical status, functional status and access. Determine the numbers and skills of available health staff. Determine the capacity and functional status of existing public health programmes, e.g. Expanded Programme on Immunisation (EPI), maternal and child health services. Determine the availability of standardised protocols, essential drugs, supplies and equipment. Determine the status of existing referral systems. Determine the status of the existing Health Information Systems (HIS); make recommendations for what should be done if none exists. Determine the capacity of existing logistics systems, especially as they relate to procurement, distribution and storage of essential drugs, vaccines and medical supplies. Consider data from other relevant sectors Nutritional status; Environmental conditions; Food and food security. When possible, use standard guidelines and forms, and include information about other sectors. Before carrying out the assessment, check with the lead health agency in the emergency to identify if a standard assessment form is being recommended, as this is usually common. If so, use the standard assessment form and process.
Background health information Identify pre-existing health problems and priorities in the disaster-affected area prior to the disaster. Ascertain local disease epidemiology. Identify pre-existing health problems and priorities in the country of origin if refugees are involved. Ascertain disease epidemiology in country of origin. Identify existing risks to health, e.g. potential epidemic diseases; Identify previous sources of health care; Determine the strengths and coverage of local public health programmes in refugees' country of origin.
240
Depending on the situation, look at specific items, such as norms and attitudes regarding gender and gender based violence, existing protection and response systems and the nature of gender based violence in the community. Initially, it might only be feasible to collect age-specific data about under fives and for those above five (<5 years and >5 years).
Sources and methods for collecting data

Good sources of information for the initial assessment can be identified according to the background information and the type of information needed. Review available health records at health facilities to determine the major causes of death and illness during the initial assessment. If records are poor or not available (e.g. destroyed by floods), establish the commonly perceived causes of mortality and morbidity by interviewing local health workers and health officials. To minimise bias, the beneficiary population can also be consulted during the initial assessment or shortly after. Ask specifically about children to identify if the beneficiaries have problems utilizing health services. This also helps involve beneficiaries in the design of health services from the outset. Selected beneficiaries should be from diverse backgrounds such as age, gender, and ethnicity. The dispersed population/camps administrator and health workers are also important sources of information. How the data will be collected during field assessment depends on the nature of the emergency and the time and resources available for the assessment including the skills of the assessment team. Data can be collected using quantitative and qualitative methods such as observation, reviewing existing records, key informant interviews, patient narratives, focus group discussions and surveys. Possible sources and methods of collecting data in emergencies are summarised in the table below. (For details about qualitative research methods, consult texts listed under references and suggested readings).
Table 6-7: Sources of information and methods of data collection for a rapid health needs assessment
Source * Information to be collected Method of data collection
Affected population
Host government authorities
Health authorities (local/central ministry of health (MOH)) Health facilities (MOH, private, NGO) Humanitarian agencies (international and local), multi-lateral agencies (e.g., UN), media, internet web sites
Background information, pre- and postdisaster community health information, environmental conditions, needs and available resources, future prospects. Background information, demographic profile of local and displaced population, needs and available resources, future prospects. Morbidity and nutritional status, reproductive health needs, environmental conditions, health policies, needs and available resources. As above for local (and perhaps displaced) populations, needs and available resources. Background information, pre-/post-disaster demographic and health status data, needs and available resources, future prospects.
Surveys, observation, mapping, interviews, focus groups. Mapping, interviews, review census and survey reports (e.g. Demographic Health Survey). Interviews, review registers, surveys, reports.
Observation, interviews, review registers, surveys, reports. Interviews, review registers, surveys (e.g., demographic health survey), situation reports.
* Refer to the previous checklist for full details.
Conducting the rapid health needs assessment

A needs assessment can be carried out by following the same logical sequence as individual patient assessments, which includes: briefly observing patients on arrival, taking their history (interviews and review existing records), physical examination, interim diagnosis, planning a follow-up assessment if necessary. Similarly, after collecting the background information, the following steps may provide a logical approach to a field assessment: 1. Preliminary observations should be done if possible when approaching the site by vehicle or aircraft. Assess the environment and extent of the disasters damage and population displacement. 2. Interview officials from the local government, the public health sector, local organisations, volunteers, health workers and the affected community (leaders of different ethnic groups, women) etc. in order to: Confirm or update background information on the health needs and local response; Identify individuals or groups of people in life-threatening situation; Whether the host government will accept intervention. 3. Review existing records at the local or national level (host country, agencies, media, and health facilities) including maps, aerial photos, census health data etc. This helps establish baseline information on demography, the displaced and the host populations health status, the existing services function and capacity (e.g. relevance of the health information system). 4. Detailed visual inspection: walk around the displaced community and surrounding areas to investigate rumours and gather valid impressions about the following: The layout of the camp or settlement (from the highest level e.g. from the top of a hill, tree, or building), the estimated number of people involved, the local infrastructure and the resources; The living conditions and access to sanitation, water supply, food supply, health services and the level of insecurity; How much normal life and the social structure has been disrupted, the affected populations coping mechanisms and any other issues of secondary priority such as reproductive health needs and mental health. Note: where the disaster affected community is widely dispersed within a host population, a walk-through tour may help provide insight on their distribution. Some areas might be more densely populated than others for various reasons. The more populated areas might have worse living conditions. 5. Rapid surveys: Conduct rapid surveys using convenience or cluster sampling of households to estimate the demographic profile, health status including immunisation status and priority health problems of the affected population. Innovative methods of sampling can be adopted to sample enough displaced persons per cluster where a disaster affected community is dispersed within a host population setting. Examples include snowballing technique (locating the first subject and enquiring for
Interviewing affected community members in Guatemala Photo: International Federation
242
information on the next subject with similar characteristics) or measuring the proportion of displaced persons among all the subjects surveyed within a cluster and projecting this number to estimate the entire disaster affected population. 6. Prepare a basis for ongoing health information: Use the assessment findings to set up a health information system. Ongoing collection and analysis of information over time will refine the findings of the initial assessment. Population surveys can be organised soon after the assessment to: Provide valid base-line information if missing; Determine the priorities for the programme (e.g. cholera prevention and control, selective feeding, measles immunisation, etc.). 7. Preliminary analysis: A timely and careful analysis of the assessment findings is necessary to provide a basis for programme planning. However, the skills or the resources to carry out a detailed analysis may not exist. Simple analysis procedures may be performed in the field, including summarising statistics, frequency tables, calculating percentages, rates, and plotting graphs. Key results must be compared to normal reference values or standards to improve understanding and conclusions regarding the disaster situation and help determine the appropriate response. For more details, please refer to the annex on data analysis. Assessment findings help guide the identification of priority actions for the acute emergency phase. Initial interventions often focus on the prevention and control of diseases of epidemic potential (such as acute respiratory infection, diarrhoea, measles, malaria, meningitis). Priority interventions can include the provision of a sufficient and safe water supply, adequate sanitation, measles immunisation, disease surveillance, and distribution of bed nets, food aid and food security, shelter as well as provision of basic clinical and reproductive health services (Minimum Initial Services Package). As the situation stabilises, the range of services can expand according to the needs of the affected population, the evolving disaster situation and the available resources. This requires a more comprehensive Health Information System. The table below summarises emergency and post-emergency interventions as recommended by the Sphere Humanitarian Charter.
Table 6-8. Priority health services in emergencies by phase
Health service Emergency phase Post-emergency phase
Child health
Curative care
Immunisation (measles, Vitamin A), Growth monitoring Curative (acute respiratory infection, diarrhoea measles, malaria, skin infections, anaemia), referral for danger signs, Basic health care (triage, outpatient, referral, inpatient, dressing/injection) Referrals Manage minor injuries Referral of emergency conditions and injuries Temporary field unit if poor access to referral hospital
Expanded Programme on Immunisation programme (measles, diphtheria, polio, whooping cough, TB) Treatment of TB and other chronic diseases (diabetes mellitus, hypertension, arthritis) Surgery for chronic conditions such as hernia or uterine prolapse
Surgery
Health service
Emergency phase
Post-emergency phase
Reproductive health
Minimum Initial Service Package (MISP):
Pharmacy
Push system for drug kits New Emergency Health Kit
Laboratory
Mental health
Initially clinical diagnosis or referral of specimens; Seek to provide tests for detection of outbreaks (cholera, dysentery or meningitis) or high drug-resistance (malaria, dysentery) Prepare plans.
Comprehensive reproductive health care: Antenatal care, emergency operations centre, prenatal care, Family planning, STI/HIV Prevention and treatment Pull system for ordering drugs Essential drugs supply Standard treatment protocols Basic laboratory investigations (malaria, helminths, haemoglobin, gram stain, sputum smear, blood sugar, HIV test). Possibly blood transfusions Community-based programme for the emotionally traumatised Ongoing surveillance Population-based surveys Periodically review and update HIS Community mobilisation for disease control activities Tertiary care for physical disability
Health Information System (HIS)
Initial assessment Establish surveillance system using simple indicators
Preventive health
Community based activities include: IEC, ORT, disease surveillance, population estimates
For more detailed descriptions about specific interventions for the acute emergency and post-emergency phase, please refer to the relevant chapters. 8. Report findings: After the analysis, write an assessment report as soon as possible about the key findings and recommendations under the following headings: The assessment methodology; The disaster situation; The affected population; The local response and capacity; The external resources needed; The recommended actions.
244
Vital assessment data or findings can be immediately transmitted via the Internet, by voice or SMS. All findings and analysis must be shared with government authorities, coordinating body and/or cluster and other agencies to ensure full coordination and preparation for a unified and prioritised set of interventions. The following information about the assessment must be clearly stated in the report: Time for monitoring trends by mapping when events happened, (e.g. before and after flight); Place for comparing different sites, (e.g., camp A and camp B, or with host population); Person for calculating age-specific and sex-specific rates to identify the population subgroups at increased risk. 9. Dissemination: Give feedback to everyone who participated or has a stake in the assessment. The report should include information about the following: The assessment; The disaster; The affected population; The local response and capacity; The external resources needed; The recommended actions. 10. Recommendation for follow up surveys: Try to indicate as early as possible, which in-depth assessments are required urgently. 11. Special considerations during response: The findings should have some bearings on future surveillance and intervention in recovery and long term rehabilitation efforts. Many health care systems are disrupted after major disasters. Facilities left standing might be only operating at a reduced capacity especially at the disasters peripheral level mainly due to the lack of resources and mismanagement. This health care gap will likely widen over time with a brain-drain of doctors and nurses from the periphery. As disasters become more frequent, the time for recovery and return to any sense of normalcy will become a bigger problem for future governments because many continue to spend large proportions of the health care budget (even 60 to 80%) on higher level of care. These challenges must be considered during the emergency response in most developing countries: how to set up emergency operations with incompetent health care systems.
Surveillance
Surveillance is defined as the ongoing, systematic collection analysis and interpretation of health data, linked with giving feedback to people at all levels of the data collection system as well as applying the information to disease prevention and control measures.
Setting up a surveillance system

A surveillance system for emergency health care should be started from the initial needs assessment. The goal of surveillance is to give timely information about health problems so that diseases and outbreaks can be detected early and health services respond more effectively. Objectives of surveillance include: Monitoring a populations health and identifying priority immediate and long-term health needs; Following disease trends for early detection and control of outbreaks; Assisting in planning and implementing health programmes;
The capacity and use of surveillance will vary according to the phase of the disaster. The following table presents differences between surveillance systems set up in the acute emergency and post-emergency phases.
Table 6-9: Surveillance systems in emergency and post-emergency phases
Emergency phase Post-emergency phase
Duration Method of data collection
1-4 months Screening initial assessment, Simple surveys, Observation by walking around Reduce mortality rates
From the first month(s) onward Regular population-based surveys, Ongoing Health Information System Detect disease outbreaks, Design and monitor programmes, Monitor quality of programmes Both passive and active collection, More quantitative Census and supplemental surveys May include lab confirmation, More in number Formal with process in place, Reportable disease list Comprehensive, Used to assess quality, For longer term health needs, Addresses less urgent issues, (Emphasises public health approach)
Main Priority
Type of data collection Defining population Size Case definition
Mostly active collection, Largely qualitative Sample survey methods Simple clinical signs and symptoms, A few common conditions Informal, Watch for measles, cholera Simple, Data needed for immediate actions
Outbreak investigation Surveillance and use of data
Indicators and sources of information

Good surveillance requires standard data collection, analysis and reporting procedures. Because the time and resources for collecting, analysing and reporting data are limited particularly in the acute emergency phase, only the most essential indicators should be selected based on practical use. The following indicators should be included in surveillance systems for all phases of an emergency: Demographic indicators estimate the total population, vulnerable groups and in-andout migration; Health Status indicators estimate mortality rate, morbidity rate and nutritional status; Programme process indicators estimate the coverage of immunisation, health services, food distribution, water and sanitation. The sources and methods of gathering information should be carefully selected. The following table defines specific indicators and potential sources of information commonly used in emergency surveillance systems.
Ensuring resources are targeted to the most vulnerable groups; Monitoring the quality of health care; Evaluating the coverage and effectiveness of programme interventions.
246
Table 6-10: Surveillance indicators and sources of information
Surveillance
Indicators
Sources of Information
Demographic
Mortality*
Total population Population structure (age, sex) Rate of migration (new arrivals, departures) Identification of vulnerable groups Births Crude Mortality Rate (CMR) Age-specific Mortality Rate (<5, >5) Cause-Specific Mortality Case Fatality Rate (CFR)
Registration records, Population census, CHW reports Volunteers Hospital death registers, Religious leaders, Community reporters (including Community Health Workers), Burial shroud distribution, Burial contractors, Graveyards, Camp administration Outpatient and admission records, Laboratories Feeding centre(s) records, Community health worker records Nutrition surveys MCH clinic records Feeding centre records Birth registers Camp administration
Morbidity 1. Routine 2. Outbreaks (daily)
Incidence rate (new cases) Prevalence rate (total existing cases) Age-/sex-specific morbidity rate Proportional morbidity rate
Nutrition (Frequent surveys while malnutrition rate is high)
Programme process
Global malnutrition rate Severe malnutrition rate Rate of weight gain/loss in MCH clinics Incidence of micro-nutrient deficiency disorders Incidence of low birth weight Average daily ration Delayed age of menarche Shelter coverage (link with incidence of ARI) Feeding centre enrolment and attendance Water and sanitation (quantity, quality, access, monitor against diarrhoea incidence) Immunisation coverage Reproductive Health (Ante natal care attendance, deliveries, family planning uptake, STI/HIV prevalence) Outpatient and inpatient attendance ORS distribution Community based mental health care (number of people reached, cases with nonspecific illnesses)
Facility records, Immunisation surveys (annual), Traditional birth attendant records
*Note: The mortality or death rate is the most important indicator of serious stress affecting a displaced population. Death rates in acute emergency situations have been known to exceed five to sixty times that of normal situations. However, measuring this critical indicator during emergency situations may be difficult because data from health facilities death registers might be incomplete; other methods of data collection must be, therefore, considered (e.g. hiring graveyard monitors, interviewing grave-diggers and shroud distributors as well as doing community surveys). Each method or source of gathering information should be evaluated for quality and reliability.
Alternatives to surveillance
Surveillance systems are often biased because they collect information passively, i.e. they mainly focus on people who use existing services. Other methods of gathering information are necessary for detecting health problems and cases occurring outside the existing facilities. Other methods of collecting information will follow.
Community-based surveillance
In community surveillance, a limited amount of health information is gathered directly from the community (e.g. new cases with a common disease). This might require training volunteers, community health volunteers and extension health workers to recognise and manage cases according to their diagnostic skills. Broad case definitions can help community outreach workers to recognise and refer all possible cases to health facilities. Qualified health workers in higher levels of the emergency health system can be trained to use more specific (but less sensitive) case definitions, which may require laboratory confirmation. This will ensure the surveillance system does not miss any person that is a probable or definite case with a communicable disease. The table below gives examples of case definitions that may be appropriate for workers at different levels of a primary health care programme.
Table 6-11: Case definitions from home to hospital
Diagnosis level of care Possible case home Probable case first level facility Definite case hospital
Malaria
Fever only
Fever + periodic shaking + chills Fever + rash
Rapid diagnostic test or Positive slide for malaria parasites Fever + rash + cough or Kopliks spots
Measles
Fever only
Disease surveillance can be improved by encouraging the use of standard case definitions for diagnosing and managing patients and recording data in health facility registers. All patient and hospital records must be monitored regularly to ensure that the recorded diagnosis accurately represents the patients condition. Note: Cases diagnosed at different levels of care should be analysed separately.
Surveys
Sometimes it is necessary to organise focused assessments to gather information that is not immediately available through the existing surveillance system. For details on surveys, refer to the next section on population surveys. Key differences between surveys and surveillance are described in the following table.
248
Table 6-12: Differences between surveys and surveillance
Surveys
Surveillance
Intermittent, focused assessments that collect population-based health data (active). Collect information on demography, morbidity, mortality, nutritional status (acute malnutrition) and programme indicators (e.g., Expanded Programme on Immunisation, reproductive health, and use of health services). May be limited to concerned agency/facility. With appropriate sampling, allows for filling of information gaps in communitylevel data. Requires more time and resources to organise, but is a one-time cost only.
Ongoing, systematic collection, analysis, and interpretation of facility-based data (mainly passive). Collect information on demography, morbidity, mortality, births, nutritional (micro-nutrient deficiencies), health services and environmental health indicators. Should involve all health agencies and facilities. Captures those who attend facility-based services, therefore not representative of all needy groups. Less costly since integrated within routine services and the existing system.
Outbreak investigation
A surveillance system should be sensitive enough to pick up the first few cases with diseases that have epidemic potential (see table below). This can be achieved by training all data collectors to recognise cases with reportable diseases. They should also be given guidelines for immediate reporting of a suspected disease outbreak. All reports should prompt immediate action by the appropriate health authorities beginning with a preliminary investigation to confirm whether there really is an outbreak. (Refer to the section on outbreak investigation for further details).
Table 6-13: Examples of diseases that can cause outbreak
Reportable diseases
Diseases of public health importance
Measles Cholera Meningitis Hepatitis Tuberculosis Yellow fever Haemorrhagic fever
Rabies Tetanus Sexually transmitted infections (gonorrhoea, syphilis, chlamydia, genital ulcer disease, chancroid) HIV/AIDS
Analysing and reporting surveillance data

For a surveillance system to be useful, the information that is gathered should be analysed and reported in a timely manner. Data analysis includes summarising data into frequency tables, calculating rates, plotting simple graphs and comparing all information with earlier information. As much data analysis as possible must be done at the field level where it can be used. This will improve the programmes effectiveness. Staff responsible for analysing and reporting surveillance data need to do the following:
Table 6-14: Example of a 2x2 Frequency Table

Cases with ARI Males Females
Age of patients
Total
Under five years Over five years Total
a c a+c
b d b+d
a+b c+d a+b+c+d
Since the main sources of information are health facilities, the most senior person at each health facility should be held responsible for performing simple data analysis (e.g. sorting and summarising data and calculating rates) and forwarding it to the health co-ordinator. At the project office level, the health co-ordinator might perform additional data analysis and interpretation before reporting the findings to the headquarter level and lead agency. The analysis and interpretation of all health information should be linked with feedback to the data collectors. Copies of the surveillance reports must also be forwarded to the district and national health offices (either on a weekly or monthly basis). when a potential disease outbreak, however, is suspected, the district health office should be notified immediately. (For more details, please refer to the section on data analysis and reporting at the end of this chapter.)
Evaluating a surveillance system

A surveillance system can be evaluated and updated from time to time. The following table describes the criteria for evaluating surveillance systems:
Table 6-15: Criteria for evaluating the function of a surveillance system
Criteria Description
Simple Representative Relevant
Data is easily collected and recorded in a logical, transparent manner. Indicators used are in line with the defined problem, e.g., use Weight-ForHeight, not Weight-For-Age to assess for acute malnutrition. Limited to relevant public health information that can and will be acted upon, e.g. prevalence of intestinal worms is not a priority indicator of health status during the acute emergency phase. In detecting any outbreak (might depend on the frequency of reporting data) Information is gathered in a standard manner (case definition, tools, procedure) and can be reproduced. Indicators should mean the same to data collectors at a particular level, e.g. the case definition for malaria is the same for all community health workers Performs repeated measurement of the same indicator to detect trends. To both the affected population and to the authorities. Can adapt to new health problems or sudden programme changes.
Timely Reliable Standardised Continuous Acceptable Flexible
Focus on mortality rates and key causes of illness; Display disease trends in form of graphs; Ensure information is passed promptly to decision-makers in a manner they can easily understand; Give feedback to the data collectors after analysing and interpreting the information. Refer to the data analysis section for further details.
250
The following indicators might also be used for evaluating the surveillance system:
Percentage of cases or deaths reported as Unknown or Other; Suitability and use of standard case definitions; Ways of disseminating findings from surveillance; Who gets and who uses the surveillance data; Procedures for making inquiries and for direct reporting of epidemics; Use of surveillance findings in decision-making and action.
Population surveys
Surveys are defined as periodic, focused assessments carried out to collect additional health data from a population. They gather information that is not routinely collected by the existing information systems (e.g. to find out if the displaced population has access to food, water, health care etc.).
Basic principles
There are two types of surveys exhaustive surveys that study the entire population and sample surveys that study a subset of the population. Exhaustive surveys might involve too much time, money, manpower and create many errors. Well-designed sample surveys can provide more valid information about the entire population than interviewing each member of the population.
The DPRK Red Cross has dispatched teams of specialists to Kangwon, South Phyongan and South Hamgyong provinces to make further assessments on the impact of the floods and monitor the distribution of relief packages to the affected people. Delegates from the International Federation are also taking part in the assessment visits.
Relief workers will encounter many situations where they might have to carry out a survey. Because surveys consume many resources (staff, time and money), relief workers might first confirm that the survey information is not available from all possible sources and that a survey is the best way of obtaining it. Surveys can investigate the entire population or only a fraction of the target population (sample surveys). Most health related surveys are sample surveys. The following table outlines the necessary steps for planning and organising surveys:
Table 6-16: Checklist for conducting surveys

Planning Organising
1. Plan the survey

Identify the health problem and its importance Determine what additional information is required to solve the identified health problem Establish why a survey is the best way of obtaining the necessary information 2. Survey design Define survey objectives, e.g., to determine the prevalence of malnutrition in children below five years List the main questions the survey should answer to achieve the survey objectives Outline methods and instruments for gathering data Estimate at least two weeks to train staff, conduct the survey and analyse data Arrange for four to five survey teams, each with a local supervisor and two interviewers Arrange for equipment and supplies: e.g. GPS, height boards, compass, computers, stationery 3. Plan how results will be analysed and reported Before the survey begins, write up the background and the survey methodology Work out the main end-results expected from the analyses in form of dummy tables Draw an outline of the survey report: section headings, tables, graphs, etc. 4. Sampling Define the population to be surveyed and their location based on the survey objectives, e.g., under-five population may be assessed for malnutrition Decide on sampling method and calculate the sample size. Cluster and systematic random samples are most common. Outline the logistical plan for sampling 5. Design the survey questionnaire Select indicators and limited number of questions according to defined objectives. Translate and back-translate the questionnaire using different persons. Keep key terms in local languages Pilot test the questionnaire, methods, equipment and analysis procedure
6. Prepare the community

Inform the community leaders about the purpose and method of the survey Get their agreement and co-operation 7. Train the supervisors and interviewers (and translators) To sample respondents as needed To keep their respondents interested during the interview To ask each question in a standard way To take measurements correctly and record data on the questionnaires Practice sessions may help identify problems in understanding questionnaire by data collectors and households. Make any corrections to the questionnaire or the training before starting the survey 8. Conduct the survey Involve community members and leaders Arrange for supervision and regular discussion Review completed questionnaires with the interviewers 9. Analyse and interpret the data Manually tabulate the data (tables, frequencies) Calculate averages, percentages, rates, etc. Graph and tabulate analysis results Nutrition measures may be tabulated and analysed using computer software such as Epi-Info (from CDC/Atlanta) Interpret results in light of other information 10. Survey report Write a survey report and present findings to and receive feedback from the community, MOH, other NGOs, and survey data collectors Incorporate data and feedback into health information system Develop recommendations and action plan from survey results and feedback (no survey without action) 11. Evaluate the survey If survey objectives were achieved, key lessons learned in the process Programme changes resulting from the survey Effectiveness of revised programme in addressing health problem and needs identified under step 1
252
Survey design
A population survey is carried out to achieve particular objectives. These will depend on the main problems affecting a displaced population. Objectives for the population survey can be selected from the following: To measure the incidence or the prevalence of a disease or health condition, such as malnutrition; To measure past events such as mortality rate during a certain period; To estimate the coverage or use of specific services such as immunisation and outpatient clinics; To identify groups at increased risk of specific conditions (vulnerable populations) that should get treatment or referral to health services; To learn about local beliefs, customs, practices etc. relating to health; To test a hypothesis (an educated guess or theory) about the link between risk factors (e.g. hook worm infection) and presence of a health condition (e.g. anaemia). After setting the objectives, define the main questions that the survey will try to answer. For example, for a population with high levels of anaemia, a survey may help determine whether the presence of anaemia is related to hookworm infection. These questions can be compiled into a questionnaire to be used for gathering the required information.
Planning the analysis and reporting

How information from a population survey will be analysed and reported needs to be determined early in the planning stage. Dummy Tables are useful for summarising survey findings, as shown below.
Figure 6-2: Examples of Dummy tables for a hookworm survey Distribution of hookworm infected people by age and sex
Age of patients Cases with hookworm infection Males Females Total
Under five years Over five years Total Distribution of hookworm-infected cases by levels of haemoglobin
Hookworm infection Haemoglobin level Present Absent Total % with hookworm
Less than 10 g% 10g % or more Total % with anaemia
Estimating sample size

Table 6-17: Estimating sample size
Estimating sample size for sample random or systematic sampling Estimating sample size for cluster sampling
N = Z2pq d2 N = size of sample Z = level of statistical certainty chosen, or confidence interval: 95 % => Z = 1.96; 90% => Z = 1.68 value of z usually rounded to 2 d = degree of accuracy desired = half the confidence interval p = estimated level/prevalence/coverage rate being investigated. (When in doubt, use 50% for maximum sample size.) q=1p Example: To calculate the largest sample within a 10% margin of error and confidence limits of 95% (z = 1.96): N = (1.96)2 (0.5) 2 = 3.84 x 0.25 = 96 0.01 (0.1) 2 Note: Confidence interval is the range of values obtained from the sample survey between which we are 95% confident that the true value in the overall population exists. The above calculations can easily be performed using the STATCALC function on Epi-Info2. Epi-Info is a public domain software package that can be used to process questionnaires, manage epidemiological databases and perform statistical calculations, including sample size calculations and data analysis.
N = 2 (Z2pq) d2 Because this type of sampling has some degree of selection bias, one should approximately double the sample size. Doubling would maintain the same degree of precision as simple random sampling: N = 2 X 96 = 192 WHO immunisation coverage survey uses a minimum of 7 subjects per cluster (7 x 30 clusters = 210, which is greater than 192). The rapid KPC 30-cluster survey (of Child Survival Projects) uses a sample size of 300 (10 per cluster) to ensure that sub-samples are large enough to obtain management type information within statistical margins adequate for making management and programme decisions. Note: A cluster sample is recommended for collecting information on condition that is common. It does not give sufficiently accurate estimates for rare conditions.
The above table shows formulas for calculating the appropriate sample sizes for different sampling methods. In general, the larger the sample, the more reliable the estimated results of the entire population will be. Therefore, the size of a selected sample should be large enough to give reliable estimates, but not so large that it wastes limited resources. A sample size of 100 to 200 randomly selected individuals from a population of up to 20,000 is usually adequate to assess for a common condition. However, a larger sample size is needed when greater accuracy is required or to investigate conditions that have a low prevalence (e.g. maternal deaths). Note: Sample size tables in standard statistics textbooks can be used to determine the actual sample size needed.
254
Sampling methods
Sampling is the selection of a specified number of persons in a population for study with the hope that they are representative, i.e. the characteristics of the sampled population (study population) are similar to the population from which it is drawn (reference population). In probability sampling, every person in the target population has the same known (and non-zero) chance of being included in the survey. It allows investigators to form conclusions about a reference population based on information collected only from a subset of the population. Probability sampling therefore enables the collection of reliable information at a minimal cost. Results from these surveys can also be compared with results of similar surveys performed in another time, place or population. The following are probability sampling methods.
Simple random sampling

Begin by constructing the sampling frame. List every sampling unit (persons, household, village, etc.) from which the sample is to be drawn; Randomly select the required number of units from the sampling frame by drawing lots or using a table of random numbers. Note: This method is more likely to produce a representative sample, but it can be expensive and difficult to make a sampling frame where a population is scattered.
Systematic sampling
Begin by randomly selecting the starting unit to fulfil statistical requirements in systematic selection; The next sampling units are systematically selected by adding a certain number n (e.g., ten, twenty and fifty depending on the sample size relative to the total population) to the starting unit. Note: A systematic sample can be drawn without an initial listing (e.g. choose from a line of people or according to the time patients enter a clinic.
Cluster sampling
Cluster sampling begins with a list of clusters (community or administrative subdivisions, e.g., sub-location, village, zone, plot etc.). For the first stage, a certain number of clusters (usually thirty) are randomly selected, based on the cumulative frequency distribution of a population. For the second stage, a specific number of sub-units (a minimum of seven) are randomly selected within each selected cluster. Determine the direction line by spinning a bottle on the ground; The starting sub-unit (e.g. household) is randomly selected by picking a random number between one and the total number of households along the direction line between the geographical centre (e.g. market, mosque, church) and the cluster boundary; Subsequent households might either be the nearest household from the entrance of the one just sampled or every nth household (where the sampling frequency might be the total number of households along the direction line, divided by the required number of subjects per cluster, (e.g., seven); In each selected household, a suitable subject (e.g., child younger than five) is sampled and examined if present. The selection continues until all the required subjects per cluster have been interviewed. Note: Consider the following points when selecting subjects in a cluster: If the suitable subject in a selected household is not available, select the next household in the same direction;
The following figure shows an example of cluster sampling.

Figure 6-3: Example of cluster sampling (WHO)16
Sampling method The estimated population of Garissa District for 1998 is 231,022 distributed among twelve divisions and eighty-four sub locations (Office of the President, Garissa District Development Plan: 1997-2001). This figure does not include the 120,000 Somali refugees living in camps at Dadaab. Sampling was done among the non-refugee population using multi-stage cluster sampling. Thirty sub locations were selected as clusters. The number of clusters selected per division was based on population distribution by division. Sub location clusters were selected randomly from each division. Seven households were chosen in each cluster by random sampling. The investigating team identified one individual in each household for recruitment into the study. Whenever possible, each cluster included one child aged two to nine years, five individuals aged ten to fortynine years and one individual fifty years or older to reflect the age distribution of the RVF and HF cases already identified in this outbreak. The target sample size was 210 individuals.
Exhaustive sampling
Sometimes, the entire population must be surveyed e.g. when investigating a disease outbreak or selecting only a group of people for the study can create a strong feeling of discrimination in the population.
Choosing a sampling method

The type of sampling method selected depends on whether a sampling framea list of individual people from which a sample is to be drawnexists. When a map with a prior census is available, subjects can be selected using simple random sampling or systematic sampling. However, a second-best cluster sampling procedure is the only option where a population has not been settled in an orderly manner (as is the case in many refugee camp settings). The most commonly used sampling methods are systematic sampling and cluster sampling. The following table describes the main differences between these two methods.
If there are not enough subjects in the cluster, the survey team for that cluster should go to the next nearest house in the next nearest cluster; If the last household has more than one suitable subject present (e.g. three children less than five-years-old), include the other children in the survey.
256
Table 6-18: Systematic sampling and cluster sampling
Systematic sampling
Cluster sampling
Ideal where shelters are arranged in an orderly manner. Requires more effort to construct a detailed list of individual subjects (from census or registration) as a sampling frame. Systematic selection may not necessarily produce a random sample. Provides more precise estimates of the reference population from a similar sample size.
Most suitable method when the site is not arranged in an orderly manner and the population is large and spread out in groups. Survey is faster because people are grouped together. Less effort because only a simple sampling frame, lists of clusters (e.g., villages) with population estimates, is needed. Potential for errors if the variable (e.g. disease) being studied is clustered within the population. Less precise, therefore, requires a larger sample. The sample size formula is used but, because of the design effect, n is multiplied by two.
Designing a questionnaire
Questionnaires are useful for collecting information that may be difficult to obtain in any other way. Although designing a questionnaire might look simple, it is in fact difficult. Whenever possible, use pre-tested questionnaires from the local or international organisations (such as the host ministry of health, WHO, Demographic and Health Survey (DHS). Develop new questions for any additional information. Expanded Programme on Immunisation information can also be referred to as a useful data collection and analysis software. Pictures are useful for illustrating questions that are difficult to state in words or for illiterate data collectors. To develop complete questionnaires, focus group discussions can be used to develop the first draft.
Key steps for designing questionnaires

The following steps outline the process for designing a questionnaire: 1. Define indicators that meet the survey objectives, including definition of cases and events; 2. Identify the easiest method for assessing each indicator; 3. Develop questions that produce the required information for each indicator; 4. Check each question against the survey objectives. Keep only those questions that provide the most essential information; 5. Ensure each question is clear, simple, short and easy to ask; 6. Decide whether to make the questions open-ended (i.e. narrative response) or closed (i.e. coded); 7. Test new questions on dummy tables (see planning the analysis and reporting section) to confirm that they will assess the selected indicators; 8. Translate the questionnaire into the local language and then translate it back to the original language to identify any mistakes; 9. Ensure there is a logical flow of questions in each section. Begin with general questions and end with the more sensitive questions; 10. Place instructions for the interviewers at the beginning of each section; 11. Provide enough space between questions for recording responses;
13. Consider all problems encountered during the pilot test of the questionnaire and make final changes.
Figure 6-4: Sample questionnaire for a childhood mortality survey
Questionnaire for childhood mortality survey
1. Are there any children less than five years of age currently living in this
household? (Yes/No) _________________________________________________ If yes, how many? ________________________________________________
2. Have any children less than five years of age in this household died during the past year?
(Yes/No) __________________________________________________________ If yes, how many? ________________________________________________
3. How old was each child at the time of death? ______________________________ 4. Did the child have any of the following during the week before death:
Fever with cough? Fever without cough? Diarrhoea? Fever with rash? Accident?
5. Which other symptoms did the child have? _______________________________

______________________
6. Which of the health problems do you think was the cause of death in your child?
__________________________________________________________________
7. Did the child visit the health post during the week before he/she died? (Yes/No)
__________________________________________________________________
Common problems with questionnaires

The following are common problems with questionnaires: Too many questionsafter questions on personal details (name, age, sex, etc.), add another ten to fifteen questions to limit the interview to a total of ten to twenty questions; Leading questionsensure questions are neutral. Do not suggest that a particular answer is correct; People are asked about events that they cannot recall. In general, the maximum recall period that can be relied upon is two weeks (except for major events such as admission to hospital or death); Interviewers are left free to interpret the answers. Use pre-coded answers or record exactly what the subject says; The questionnaire is constructed in one language but administered in another; Interviewees might be concerned about the nature of the questions and who will be informed of their responses. A full explanation about the survey should be given prior to the interview with strong reassurances about the confidentiality of their participation and use of the information.
12. Pilot test (try out) the questionnaire and other survey instruments (e.g. weighing scales, tape measures) in an area that is not to be surveyed. Check that no essential information has been left out and the interview is short (less than twenty minutes);
258
Training and supervising interviewers
Train all interviewers to administer the revised questionnaire in a standard way otherwise they might influence how interviewees respond to the questions. They should be trained to ask questions in a neutral way and refrain from giving advice. Even if samples are carefully selected, surveys results can still mislead. Interviewers should be closely supervised to prevent the following: Non-response bias can occur if a high proportion of the sample population is missing or did not answer the questions. For example, a survey done during the day might miss young men and women who have gone to work. People might not be willing to answer sensitive questions. Non-response bias can be minimised by the following: Ensuring that at least 80% of the original sample population responds during the survey; Following up all non-responders at least once;
In many cultures it is important that women interview women and men interview men particularly for sensitive subjects like family planning, STDs and HIV/AIDS; Observer error can arise in commonly occurring inaccuracies because the interviewers recorded faulty measurements rather than because of faulty instruments or unreliable subjects. Observer errors can be reduced by the following: Making all interviewers sign their names on each questionnaire they administer; Checking that interviewers follow standard guidelines when taking and recording measurements; Checking instruments daily and adjusting the zero reading on weighing scales.
Table 6-19: General outline for a survey report
Outline for a full survey report
Introduction Purpose of survey Survey area Dates of survey Methodology Indicators Sampling frame Questionnaire used
Survey results Highlights Graphs with charts and tables Conclusions and recommendations Significant findings Problem areas Potential actions Further investigations
Analysing and reporting survey findings

After carrying out the survey, the information in the completed questionnaires must be processed and analysed to be meaningful. It might be transferred into dummy tables and simple calculations performed in the field by hand or with the help of a pocket calculator. Further analysis might only be possible at the project office and the ministry of health levels. In the early phase of a disaster, vital data and information must be transmitted immediately because the next day things might look different. For the best possible common analysis and coordinated and prioritised action, information sharing with all agencies, cluster and government is absolutely vital. The following table shows a general outline of a report to present the data to programme decision-makers in a way that they can understand. A shorter report focusing on the survey results and recommendations from the survey must be sent to the affected community.
In May 2002, thousands of Liberians were forced to flee as a result of clashes between government forces and dissidents. Nearly 70,000 people were displaced including Internally Displaced Persons (IDPs) previously living in the camps. This last wave of displacement created additional needs for shelter, health, sanitation and food aid. All displaced persons were distributed in fourteen camps and sites located in Montserrado and Bong Counties. The full extent of the mortality and morbidity status in all camps following the forced migration in May was not well known. Nutritional status for at risk populations in all the camps and sites, particularly for the under fives, was not assessed. A baseline assessment was organised to determine the magnitude of the problem, the existing resources and the existing gaps so that all problems could be tackled. Methods Survey population Eleven camps in Montserrado and Bong counties were included in the assessment. Five of the camps were located in the Montserrado County with an estimated population of 77,000, while the remaining six camps in Bong county had an estimated population of 39,000. Sampling The sampling method selected for the mortality and nutritional assessment was systematic random sampling. A minimum sample size for the mortality survey was calculated based on the June population estimate and a sampling interval of one in ten houses was defined. Deaths since interviewees arrived to the camp were recorded and the month and year of death identified using the events calendar. Crude Mortality Rate and Under Five Mortality Rates were calculated for the month of July. Survey questionnaire Age: the age of a child under five was noted in months and determined as accurately as possible using a calendar of local events. For adults, age was registered in years. Anthropometric assessment: Mid Upper Arm Circumference (MUAC) was measured at the midpoint between the shoulder and elbow of the left arm while the arm was relaxed. MUAC is expressed to the nearest millimetre. Measles vaccination: the mother was asked if the child had been vaccinated against measles and was asked to show the vaccination card or other official document proving the child had been vaccinated. Two data entries were recorded: yes proven by vaccination card or no if a vaccination card could not be produced. Reproductive health care: any pregnant woman found in a selected house was asked if she had visited a health clinic for her pregnancy since her arrival to the camp. Any woman that had given birth since arriving at the camp was asked if she had gone with her child to the clinic to be checked within the first ten days of her delivery. Mortality: mortality data was collected from interviews with the head from all the selected households including households with no children aged between six to fiftynine months. Data included the number of people living in the household and number of people that had died in the household between their arrival to the camp and the date of the assessment. Cause of death: inquiries were made about the presumed cause of each death, reported and registered as stated by the respondent.
Case study: health survey among an internally displaced population in Liberia22 Background
260
Data collection Four data collection questionnaires for demographic, mortality and nutritional data and a fourth for health services assessment were made in consultation with the International Rescue Committee health team in Liberia. The survey instrument was pre-tested in habitats and modifications done accordingly. Five three-member teams administered the questionnaires after attending four-day training for the survey. A team leader was appointed for each team to scrutinise the record forms for errors and omissions, condense key information gathered and address any problems the other team members might encounter during data collection. Before initiating data collection in each camp, one team met with key informants to draw a map of the camp that identified how displaced people were distributed at the time of the survey. Once all blocks in the camp were identified, they were divided among the teams. Each team met with block leaders or members to establish each blocks limits. Once established, the first household of a block was selected using random numbers. After the first house was selected, every tenth house was visited until the block was completed. Houses selected for the survey were grouped into four categories: Under construction with no one living in them; Constructed but unoccupied; Constructed and occupied with occupants present at the time of the visit; Constructed and occupied with occupants not present at the moment of the visit. Whenever a house was found under construction or unoccupied the next house in the sampling interval was visited. When occupants where not present at the time of the visits they were revisited twice. If two or more eligible children for measurement were found, all of them were measured. Whenever children were not found in a house, the following house in the sampling interval was visited. At each house, interviewers asked about basic demographic information, water source and mortality for the data. A local calendar of events was used to determine household members ages and dates of death. Mid-upper arm circumference (MUAC) measurement was performed using a standard measuring tape. When a childs MUAC child was found to be below 135, the child was referred to the clinic for further assessment. One team collected health services data on morbidity and data from existing registers in the health facilities. The team leader and interviewers held two separate focus group interviews with clients at the health facility: one with women and another with men to identify better the key health priorities and possible solutions as well as the resources needed to address them. Data processing and analysis Demographic, mortality and nutritional data was condensed in a spreadsheet designed for this purpose. Dummy tables were constructed beforehand for the analysis of each of the variables. Appropriate biostatistical and epidemiological programmes were used to process and analyse the data. Health services data was condensed in a spreadsheet and dummy tables were constructed beforehand to facilitate analysis. Results The study found crude and under-five mortality rates of 3.0 and 8.2, respectively, in Mont Serrado County. The major causes of death were diarrhoea (18%), febrile illness (11%), and respiratory infection (10%). The main causes of morbidity were malaria (43%) and ARI (18%). For children under five, 8.2% had moderate malnutrition when measured using mid-upper arm circumference as an indicator of malnutrition.
IDPs in the Liberia camps are facing a critical health situation: crude mortality rates are two times over the emergency threshold. The mortality rates for children under five years old are three times over the emergency threshold.
Rapid clinic-based fever survey for malaria

Clinic-based fever surveys generate information about the prevalence of parasitaemia among symptomatic patients who seek care at health facilities.17 Using this approach, the proportion of febrile patients with malaria parasites can be quickly determined in emergency settings. The results of a clinic-based survey on fever at health facilities are presented below.
Conducting the survey

Table 6-20: Key steps in rapid clinic-based fever survey for malaria
Choose either microscopy or Rapid Diagnostic Tests (RDT) as a diagnostic method. RDTs give quick and reliable results, but may be of poor quality; Using proper staining techniques, microscopy remains the gold standard for malaria diagnosis. Ensure that sufficient staff is available and trained in the selected method. Using RDTs or microscopy, test all febrile patients attending selected high-volume, outpatient clinics. Up to 100 patients should be tested. Conduct all testing within a short time period, preferably one to three days. Record the total number of outpatients seen at the clinic while the survey is being done. Record the age, sex, pregnancy status and place of origin of tested patients. Repeat the survey weekly during the acute phase and monitor trends at least once every month.
Analysis and interpretation of survey results

Data analysis and interpretation can yield important information about malaria transmission. Using the survey data, calculate the total number of patients at the clinic and fever cases. Determine the proportion of fever cases among all outpatients. For the purpose of a rapid survey, fever should be defined as patients with fever or a history of fever over the past forty-eight hours. Calculate the proportion of confirmed malaria among those with fever. Patients have confirmed malaria if there is a presence of malaria parasites in their blood. This proportion can be represented as the parasite rate, slide-positivity rate, or positive Rapid Diagnostic Tests rate. For ease of interpretation, results can be presented by week and age group. Data can be interpreted as follows for the host and displaced populations.
Host population
Age group Proportion of parasites in fever cases Interpretation of malaria transmission
All children and adults Children under ten years
Less than 10% More than 50%
Low transmission High transmission
Conclusions
262
Children over ten years and adults
More than 50%
Possible onset of outbreak
Displaced population
Among the displaced population, interpretation of survey results is more complex. Similar results might suggest different situations, particularly during the first few weeks of settlement. The following table outlines possible interpretations for high proportions of positive malaria among febrile members of a displaced population.
Scenario Interpretation of malaria transmission
High local malaria transmission
If incoming population is non-immune, high morbidity and mortality rates will be recorded in all age groups. If incoming population is semi-immune, children and pregnant women are most at risk, showing high morbidity and mortality rates.
High malaria endemic in area of origin
If the incoming population comes from an area where malaria is often found but the local area is not receptive to malaria transmission, then over time there will be fewer and fewer malaria cases in the health care facilities, as only those who brought it from their home area will have it. However, if the area where people are relocated to is receptive to the type of malaria the incoming population is bringing with them, over time an epidemic may start.
High malaria endemic going to another area with high malaria endemic Travelled through high malaria endemic areas
The immunity acquired in the area of origin might not have the same protective capacity in the new area; the plasmodia are sometimes slightly different. As above.
Source: WHO
Community cross-sectional malaria prevalence survey

The population survey approach can be applied when information about the prevalence of malaria in a displaced population is needed. The purpose of the community crosssectional malaria prevalence survey documents the baseline prevalence of malaria among an affected population; identifies groups (e.g. children under five) at particular risk for malaria; determines information about treatment-seeking behaviour; and uses this information to guide and monitor interventions. Information collected from community surveys can help distinguish between low (well under 10% of the population is infected) and medium/high malaria endemic (10% or more of the population is infected).17 Generally, these rapid surveys take three to six days to complete.

Steps in conducting the survey are outlined here. Define the study population. See the population survey section above; Determine the sample size. See the population survey section above; Choose a sampling method. See the population survey section above; Recruit and train survey team. The team should include a community member to explain the procedure and obtain informed consent from households; one person to
Ensure adequate supplies. Essential supplies include diagnostic equipment (e.g. RDT or microscope and slides for smears), lancets, rubber gloves, cotton wool, alcohol for swabbing skin, clinic referral forms, consent and record forms, pens and pencils, anti-malarial drugs for treatment (if administered on the spot) and, if needed, funds to pay for transport to the clinic. Community cross-sectional malaria prevalence surveys should be repeated in different seasons because malaria transmission varies between dry and wet seasons. For areas where high transmission exists all year, the survey can be repeated several weeks after the start of interventions to monitor their impact. Finally in areas with annual peaks in transmission, it will be helpful to repeat the survey after twelve months to assess effectiveness of interventions.
Analysis and interpretation of survey results

At least the following two indicators should be calculated and interpreted from the collected data: Prevalence of malaria can be calculated by determining the proportion of people from the entire population with fever or history of fever and who have a positive RDT result. The data should be analysed by age, sex and pregnancy status. Special attention should be paid to prevalence among children under age five. Malaria Endemicity can be determined by calculating the proportion of the entire population with a positive RDT (or slide result) or the proportion of the entire population who have splenomegaly. The following table offers interpretations of these calculations.17 Table 6-21: Classification of malaria endemicity
Spleen rate Malaria endemicity Under five years Over five years Parasite rate Under five years Over five years
Meso- and hyperendemic Hypoendemic
Very high
Low or zero
High (>50%), but lower than spleen rate
Low
Low (<10%)
10%, always higher than spleen rate
Software programmes such as Epi-Info can be used for further data analyses.
register participants, record auxiliary temperature, the Rapid Diagnostic Tests (RDT) result and information on treatment-seeking behaviour; one person to take auxiliary temperature and conduct the RDT or make a malaria smear on a slide; one person to check for presence of an enlarged spleen; one person to administer treatment and advise about danger signs and follow-up;8
264
Table 6-22: Distinguishing between rapid clinic-based fever survey and community cross-sectional malaria prevalence survey
Rapid clinic-based fever survey Why To gather quickly information about malaria prevalence among symptomatic patients seeking care at health facilities.
Community cross-sectional prevalence survey To document the baseline prevalence of malaria among an affected population in the community. To identify groups (e.g. children under five) at particular risk from malaria.
What
Determines the proportion of treatment-seeking febrile patients with malaria. Ongoing, continuous
Used to determine levels of malaria transmission at the community and sub-population level. Periodic, intermittent. Typically conducted at baseline (e.g. before implementing intervention) and used to guide and monitor interventions. Use population survey methods by defining and selecting the population, determining the sample size and sampling methods, conducting the survey and testing participants in the community. More costly and requires more personnel. Requires a five member team (see above). Malaria is meso- or hyper-endemic if a) spleen rate among under 5 years is very high or among over 5 years is low or zero or b) parasite rate among under 5 years is high (>50%) or among over 5 years is low. Malaria is hypoendemic if the overall spleen rate in the population is low (<10%) or the overall parasite rate is low ( 10%) and higher than the spleen rate.
When
How
Using rapid diagnostic tests or microscopy, test all febrile patients attending selected high-volume, outpatient clinics. Up to 100 patients should be tested. Cheap and requires less personnel. Can be conducted by clinic staff and administrators. Among all children and adults, if proportion of fever cases with parasites is less than 10%, interpret it as low malaria transmission. Among children less than 10 years, if proportion of fever cases with parasites is more than 50%, then interpret as high malaria transmission. Among children over 10 years and adults, if proportion of fever cases with parasites is more than 50%, then consider outbreak onset.
Cost Who
Interpretation For host population:
Disease outbreaks or epidemics occur when many people in a community or region develop a similar illnessin excess of normal expectationsthrough a common source or carrier. An outbreak can be declared following the detection of a single case in a nonendemic area (such as cholera or measles) or after the number of reported cases reach the threshold incidence rate of a particular disease (e.g., threshold for meningitis is fifteen cases per 100,000 people in a two-week period). Please refer to the Control of Communicable Diseases chapter for more details about epidemic thresholds.
Objectives of investigating an outbreak

The following may be defined as objectives for investigating an outbreak: To identify the cause(s) and risk factors for the disease; To identify the appropriate prevention and control measures that will reduce the impact of the disease; To determine the extent of the disease; To provide a foundation for developing public health policy.
Key steps
Table 6-23: Checklist for an outbreak investigation
Key step Description
1. Notify and coordinate with the host and international health authorities. 2. Confirm the outbreak.
Provide essential information on the affected sites, the time period, the frequency and profile of cases, the clinical presentation and disease outcome, a possible diagnosis and suspected source of infection; Be sure to coordinate efforts with the national health authorities as well as inter-governmental investigations (e.g. WHO). Define a case and count the number of reported cases (the numerator): Is the disease known? Are the causes partially understood? Define the denominator: What is the population at risk of developing the disease? Calculate the attack rates; Review previous levels of disease and local knowledge of disease outbreaks. Graph reported dates of disease onset for all cases to establish the timing (incubation period) and the source of disease (single or multiple sources); Map the residence of all reported cases to identify the most affected areas and the direction of the disease spread (see spot map in the data analysis section); Calculate the age- and sex- specific rates to identify who is most vulnerable; Collect population data for the communities at risk (more denominators). Look for links or interaction between relevant factors (e.g. floods increasing the Aedes mosquito population and reducing access to health care resulting in an outbreak of dengue fever).
3. Describe the outbreak in terms of time, place and person.
4. Analyse what caused the outbreak.
Outbreak investigation
266
5. If necessary, conduct additional studies.
Interview cases with disease and non-cases to identify possible sources and methods of disease transmission (common source or person-to-person); Determine the proportion of cases and non-cases that had possible exposure to infection; Identify important differences between the cases and non-cases to define the individuals/groups at increased risk of contracting the disease; Collect specimens from cases and non-cases for laboratory investigation. Investigate for vectors, faecal contamination and toxic chemicals.
6. Assess the environment, if necessary, based on the analysis of the outbreak. 7. Initiate prevention and control strategies.
Source reductions (treat cases and carriers, isolate cases, and control animal reservoirs); Prevent transmission (health education, personal and environmental hygiene, vector control, restrict movements); Protect vulnerable people (immunisation, chemo-prophylactics, personal protection, and nutrition); Continue surveillance: maintain routine reporting, follow-up suspects, and set up special surveillance for new cases. The causative agent and probable routes of transmission; Description of the trend in the disease outbreak, the geographic distribution and the clinical presentation among cases; The reason for the outbreak; Disease control measures that were introduced; Recommendations for prevention of future outbreaks.
8. Prepare a report on the outbreak that covers the following points:
Even though different disease outbreaks might occur, there are key steps to be carried out in most outbreak investigations. These steps are summarised in the table above. The order of the steps might depend on the nature of the outbreak and the existing knowledge about the disease. For example: in suspected cholera outbreaks, appropriate disease control measures must be initiated at the beginning before identifying the cause or risk factors for disease. When investigating an outbreak, the first step is to confirm that there really is an outbreak. A local public health team might be capable of doing this and sometimes even identifying the possible causes and risk factors. The most effective ways of controlling the spread of disease should be initiated as soon as possible. Sometimes, there is not enough information for identifying either the cause of the outbreak or the appropriate control measures. If the outbreak seems to be spreading and causing many deaths rather than phasing out naturally, a special team of investigators (e.g. epidemiologists, entomologists, microbiologists etc.) might be invited to support the local team. The epidemiologists might help organise a case-control study that compares risk factors among people with and those without the disease. At the end of the investigation, a report should be written and shared with all concerned.
Carrying out the investigation

Because investigating a disease outbreak involves many people, it needs careful planning, organisation and supervision. Key procedures must be followed. If the disease is spreading rapidly, speedy carrying out of the investigation may be critical.
Planning the investigation

Consider access to the site and the willingness of the community to help the investigation; Consider the local climate, the daily family activities and the migration patterns when scheduling the time for data collection; Design appropriate questionnaires based on how the information will be analysed; Train interviewers to ask questions in a standard way and to practice using the surveys questionnaires and equipment before the investigation; Arrange for translators and chaperones to be present during interviews if needed; Standardise the sequence of data collection procedures. Most investigations involve epidemiological, clinical, and laboratory procedures. Data should be collected as follows: Collect information directly from the affected person or family member where necessary; Perform a physical examination after the interview; Finally, collect any required laboratory specimens (blood, stool etc.). Preserve all laboratory specimens appropriately.
Organising the investigation

Involve local authorities for security clearance, publicity and introduce the investigation team to the affected community; Sketch an organisational chart that shows the lines of authority, the roles of different teams and the link between functions; Remember, no survey without services. Prepare incentives to maintain community participation during the investigation. This might include providing some medical services or essential drugs.
Supervising the investigation

Supervise field workers by checking how they conduct interviews during the survey and their accuracy in recording data; Evaluate the collection and processing of laboratory specimen for quality control; Carry out and check data entry daily and perform simple calculations; Conduct frequent staff meetings to identify and address any problems. At the end of the investigation, organise a final meeting with all supporters and participants to thank and give them feedback about the investigation and recommend any long-term measures to prevent and control future outbreaks.
268
Case study: cholera outbreak investigation after the civil conflict in Liberia20
In June 2003 at the height of Liberias civil war as rebel forces approached the capital city of Monrovia, an estimated 300,000 Internally Displaced Persons (IDPs) settled in private homes with family members, public buildings and in other sites. Because of fighting between June and July, the normal collection of health data by the Liberian Ministry of Health (MoH) was interrupted. In June, cases of cholera were confirmed by international nongovernmental organisations. To estimate the magnitude of the outbreak in August, WHO conducted a retrospective review of data collected by health organisations between June and August 2003 but did make a report to MoH. Additional data was collected from an emergency surveillance system that began operation on August 25. During the week ending October 20, a total of 1,252 cases of suspected cholera were reported (WHO, MoH, unpublished data, 2003). The epidemic began in June (see the figure below) and was associated temporally with increased fighting and the movement of IDPs. Because cholera transmission was probably attributable to an acute shortage of clean water, poor sanitation and crowded living conditions, international and Liberian organisations attempted to supply IDP settlements with sufficient potable water and began chlorinating wells. Cases most closely approximating the standard WHO recommended case definition for use in cholera outbreaks (i.e. acute watery diarrhoea in a person aged > five years) were included in retrospective case counts. After August 25, the majority of facilities that reported data to the emergency surveillance system used a case definition that included acute watery diarrhoea in children aged two to four years. In June, the number of persons treated for cholera increased from forty-nine to 426 per week. Between June 2 to September 22, of an estimated one million permanent residents and 172,000 IDPs in Monrovia (1), 16,969 (1.4%) persons sought medical care for an illness consistent with the surveillance case definition for cholera. The number of persons treated for cholera increased sharply in early June and stool cultures confirmed the presence of Vibrio cholerae O1; the case-fatality ratio in cholera-treatment centres was <1%. The number of persons treated per week peaked in mid-July at 935, declined to 387 in the last week in July, and increased again to 2,352 between September 16 to-22, the last week for which data was available. V. cholerae O1 was isolated in the laboratory of St. Joseph Catholic Hospital in Monrovia from stool specimens obtained from six patients during June 9 to 13; Three choleratreatment centres operated by Mdecins sans Frontires (MSF) reported that during June 2 to September 15, out of 4,746 hospitalized patients with illnesses consistent with a diagnosis of cholera, 37 (0.8%) patients died. During this period, 3,073 (64.8%) hospitalized patients had severe dehydration. Data from the cholera treatment centre operated at JFK Hospital by MSF Belgium were used to compare the outbreak in 2003 with the number of reported cholera cases in previous years. This centre, unlike other health facilities that provided services in Monrovia during the 2003 outbreak, has treated cholera patients for the previous four years. From June to August, a total of 2,648 cholera patients were treated in this facility, compared with 450 to 655 patients during comparable periods in the previous four years.
Objectives of data analysis

The following are objectives of analysing data: To identify the possible root causes of problems; To investigate further to verify the actual causes of a problem; To define needs that has not been met; To develop an action plan for dealing with problems; To improve the quality of programmes. All data collected by whichever meansroutine information system, surveillance systems, an outbreak investigation or a survey needs to be processed, analysed and presented in a form that decision-makers can easily understand. The methods, time and resources needed to analyse data should be planned in advance. Projects can save time and effort by analysing only the priority problems being tackled by the project. Analysing data can be simple and straightforward. Most of it can be done with a pocket calculator. Basic data analysis includes the following steps: Sorting the data records; Performing summary statistics (for numerical data); Summarising data into frequency tables; Calculating percentages and rates. Where staff members lack the knowledge and skills for analysing data, on-the-job training should be organised with follow-up supervision. Where computers and staff with computer skills are available, Expanded Programme on Immunisation INFO see www.who.int/entity/chp/steps/resources/EpiInfo/en/index.html a public domain software might be used to set up questionnaires, store databases, perform basic and advanced data analysis and print results. Results of data analysis should be carefully interpreted and presented as tables and graphs, which are easier to understand.
Basic data analysis and interpretation

Data should be analysed and interpreted in a logical sequence. The following steps might be used to analyse data in the field:
Define the major end-results expected from the analysis

These are mortality and morbidity rates, coverage and access to food ration, water supply and sanitation. Note: This step may have been done when planning for the data collection.
Process all data

Data on people attending health facilities or feeding centres is usually recorded directly from individual patient cards into facility registers on a daily basis. Data in these registers will usually be summarised as shown in the following table:
Appendix: Data analysis and presentation
270
Table 6-24: Summary sheet for ARI survey data on 200 displaced people in camp A
Record No.
Date
Zone
Age
Sex
Weight
ARI infection
1 2 3 4 5 6 7 8 9 10 . . 200 TOTAL
1/1 1/1 1/1 1/1 2/1 2/1 2/1 2/1 2/1 3/1 . . 7/1
R Q S Q Q P P Q R R . . S
4 6 8 3 4 4 11 2 2 1 . . 3
F F F F M M F M M F . . F
20 25 30 15 15 20 40 12 8 7 . . 10
+ + + + + . . +
Data from surveys is usually received in individual questionnaire forms. These questionnaires must be sorted by record number and date from the earliest to the most recent. The information must be recorded onto a summary sheet similar to the one shown above. Note: All data received should be inspected for inconsistencies or for missing data and appropriate actions should be taken (e.g. verify, omit, etc.).
Analyse categorical data.

Categorical data such as age, sex, occupation, location, etc. must first be tabulated as follows: Hand-tally data from the above registers or summary sheets of observations into the corresponding box as shown below. Ensure that no observation is recorded more than once and that all the observations are contained in the table.
Table 6-25: Examples of two-way tally sheets
Example of a two-way tally sheet showing age and gender Male Female Total Example of a two-way tally sheet showing ARI cases by age ARI + ARI Total
<5 >5 Total
\\\\\ \\\\\ \\\\\ \\\\\ \\\\\ 25
\\\\\ \\\\\ \ \\\\\ \\\\\ \\\\ 25
26 24 50
<5 >5 Total
\\\\\ \\\\\ \\\\\ 15
\\\\\ \\\\\ \\\\\ \\\\\ \\\\\ \\\\ \\\\\ 35
30 20 50
Table 6-26: Examples of one-way and two-way frequency tables

A one-way frequency table: Distribution of ARI cases by location CAMP P Q R S Total A two-way table: Distribution of ARI cases by age and sex No. of ARI cases Male Female Total
No. of ARI cases
10
30
70
40
150
< 5 years > 5 years Total
50 10 60
50 40 90
100 50 150
Calculate percentage of observations: Divide the count in each cell by the grand total and multiply the result by 100. Percentages might be expressed to one decimal point or rounded to a whole number.
Table 6-27: Example of calculating percentage of observations
ARI Male No. % ARI Female No. % ARI Total No. %
< 5 years > 5 years Total
50 10 60
33.3 6.7 40.0
50 40 90
33.3 26.7 60.0
100 50 150
66.6 33.4 100
Tables of categorical data can be used later for comparison with other data for the presentation of graphs or to carry out statistical tests. Compare frequencies and percentages as follows: By person: compare frequency tables for gender/age (e.g. male with female, under-five with total population, etc.); By place: compare frequency tables for different camps or population settlements; By time: compare frequency tables with baseline or the previous months frequencies to follow trends. For all comparisons, ensure the population size and structures in the frequency table are similar. For example, the above results of Acute Respiratory Infection (ARI) distribution in one camp can be compared to results of another camp whose population size and structure is similar otherwise the conclusions drawn might not be valid. For details about how to compare populations with a different structure or size, please refer to the standard statistical texts. These comparisons can also be illustrated with graphs. Information about the one-way frequency table is better understood on a spot map that shows where the disease distribution is greatest. A spot map is easily created by pushing pins on a map of the study areas. See presenting data for an example of a spot map.
Classify data into frequency tables: Transfer total counts from tally sheets into corresponding cells (intersection of a row and a column) of empty but labelled dummy tables which should be prepared during the planning stage. These might be one-way or two-way frequency tables. One-way tables might be appropriate for classifying data by PLACE. Two-way tables might be used to classify data by PERSON (age and gender or other characteristics).
272
Analyse numerical data
(e.g., age, weight, height, haemoglobin levels, etc.) Descriptive Analysis: Summarise data by defining the following: Range-scan the data set in each category. The range is the difference in values between the lowest and the highest observed values; Median-sort the data in each category from the lowest to the highest value and note the middle value that divides the data set into two equal halves. Note: The median is referred to as the mean when the data is biased or tends to lie in one direction; Modescan the data set to identify the most common observation; Mean-calculate the mean also known as average by summing all the data in a category (e.g. birth weight) and dividing the total sum by the number of observations; Percentage-define the proportion of subjects above or below particular data categories.
Table 6-28: Examples of summary statistics
Data set for ages of 15 people with ARI infection

Original data set Sort data set: Age range: Median age: Mode: Mean age: Percentage: {2,21,3,1,4,2, 8,1,6,11,10} can begin with a disorganised data {1, 1, 2, 2, 2, 2, 3, 3, 4, 4, 6, 8, 10, 11, 21} 1 21 years 3 years 2 years (1+ 2 + 3 + 3. + 10 + 11)/11 = 80/15 = 5.3 years 10/15 = 66.7% of people with ARI are below the mean age of 5.3 years set
Standard Deviation (SD) describes the scatter of observations around their mean. A large SD implies a wide scatter in the observed data while a small SD implies a narrow scatter with little difference between the observations. In emergencies, the SD is commonly used to estimate the prevalence of malnutrition or to show the normal range of laboratory tests. For details on calculating standard deviation, please refer to standard statistical texts.
Figure 6-5: Applying standard deviation
After a nutritional survey of displaced children, the Weight-For-Height (WFH) data for all children surveyed might be plotted on a graph and compared with WFH data for a standard population of children. The fraction of displaced children with WFH more than 2 SD below the mean standard WFH is judged as wasted (shaded area under the curve for displaced children).
Table 6-29: Population distribution by age and gender

Age No. of males No. of females Total
< 5 years 5-14 years 5-14 years 50+ years Total
20 45 45 10 90
20 35 35 10 110
40 80 80 20 200
Analyse indicators
Calculate rates, ratios and proportions: Data for numerators and denominators for selected indicators can be obtained from frequency tables developed according to the previous steps. Calculate the values of each indicator (disease incidence, mortality etc.) as demonstrated earlier in this chapter. Compare selected indicators: Ensure the indicators being compared represent a similar population size and time period. By time: to follow trends, compare indicators (incidence, mortality etc.) with baseline values or those from the previous week or month; By place: to compare indicators for several locations or settlements; By person: to compare indicators for a sub-group of the population (e.g. underfive mortality) with those of the total population or if data is available, compare indicators for two sub-groups (e.g. morbidity of male and female or of two different ethnic groups in the camp).
Interpret data
Consider normal reference values or targets when interpreting health indicators. Follow trends to determine if the situation is improving. Demographic indicators-The following table lists the population structure of a typical developing country in percentages and ratios. The percentages can be applied to a total population estimate to define the estimates of population sub-groups when data about the displaced population is lacking or unreliable. Depending on the information in the demographic table available, the vulnerable population can also be estimated (e.g. all pregnant women, children less than five etc.). The population pyramid might also be drawn to display the age and sex compositions of a population (see the table and figure below).
Classify numerical data into frequency tables: Frequency tables can be used to classify numerical data under suitable class intervals. The class intervals in dummy tables might have to be revised after data collection according to the observed range of data. For example, dummy tables for a survey might have been drawn for two class intervals for age (< 5 years and > 5 years), but after the survey, the analysts might feel that they can draw better conclusions by classifying age as < 5, 5-14, 15-49, 50+ years.
274
Table 6-30: Population structure of a developing country
Population composition
Total population Infants (0-1 year) Children 0-5 years Sex ratio Women of child-bearing age Pregnant women Expected births
100% 4% 18% 1:1 24% 5% 4.4 cent
Figure 6-6: Example of a population pyramid
Health status indicators: Ensure the indicators being compared represent a similar population size and time period. Mortality: The table below shows how Crude Mortality Rates (CMRs) can be used to assess the status of an emergency situation. A CMR >1 death/10,000/day implies an acute emergency situation. The crude mortality rate of displaced populations is expected to fall below 1.0 deaths/10,000/day within four to six weeks after starting a basic support programme that has provided sufficient food and water, sanitation and health care. For well run relief programmes, CMR should not exceed 1.5 times those of the host population. The baseline CMR from the initial assessment might be compared later with other CMRs to determine the effectiveness of the relief efforts. Note: Cut-off values for the under-fives CMR are almost double the CMR for the whole population.
Table 6-31: Classifying emergency situations by crude mortality rates

Benchmarks for crude mortality rate
Indication Total Population CMR 0.3-0.5 /10,000/ day CMR > 1.0 / 10,000/ day CMR > 2.0 / 10,000/ day Age under Five Years CMR 0.7-1.0 /10,000 / day CMR > 2.0 /10,000 / day CMR > 4.0/10,000 / day Normal for stable situations in developing countries. Very serious situation. Must be investigated. Emergency out of control. Demands immediate actions.
Morbidity: Determine the importance of identified diseases in the following terms: How common each disease is and what the risk factors are; Whether the condition is potentially life-threatening or disabling (severity); Whether the control measures being implemented locally are effective for reducing the disease incidence, prevalence, severity or death from the disease; Whether the existing disease surveillance system is capable of detecting and monitoring the disease or if new indicators for the disease should be added. Nutritional status: A displaced populations nutritional status might be projected from the nutritional status of children less than five years of age. Two types of indicators can be used: Clinical indicators of malnutrition include detection of oedema (excess fluid in tissues of lower extremities), skin changes (scaling, baggy skin), hair changes or signs of micronutrient deficiency disorders. Clinical indicators must be interpreted against anthropometric indicators. Note: The presence of oedema indicates severe malnutrition regardless of the Weight-for-Height (WFH) indicator. Anthropometric indicators are based on measurements of age, sex, weight and height. There are several anthropometric indicators, but the ones most commonly used for measuring malnutrition in children are WFH and Mid-Upper-Arm-Circumference (MUAC). The following table shows the MUAC and WFH cut off values for global and severe acute malnutrition:
Table 6-32: Cut-off values for MUAC and WFH for acute malnutrition
WFH MUAC Percentile Z-score
Severe malnutrition Global malnutrition
11.5 12.5
<70 <80
< -3 SD < -2 SD
Take are when interpreting findings from anthropometric surveys. WFH might be interpreted as a percentile, median or Z-score. For developing countries with lower normal nutritional intake levels, up to 5% of children might have a Z-score below -2 Standard Deviation (SD) when compared to the reference population. Thus, relief organisations should consider that a nutritional emergency exists if more than 8% of children sampled have a Z-score below -2 SD. Finding even as few as 1% of children with a Z-score below -3 SD indicates the need for immediate nutrition interventions. Note: See the nutrition chapter for details about calculating WFH.
276
After determining the global and severe malnutrition rates for a displaced population, it is essential to interpret these rates against the following factors: Morbidity and mortality rates for children under five; Time of year (e.g. harvest or planting season); Food availability and consumption; Trends in food security.
Programme process indicators

Identify any gaps in coverage and quality of services provided (food and water supplies, sanitation, immunisation, health services, etc.) by comparing the actual values of the process indicators to pre-defined targets or standards. The table below identifies where the practices and/or coverage of immunisation fall short of what would be desired or expected.
Table 6-33: Comparing standards to current practices
Actual practice and coverage
Set standard
Conclusion
80% access to immunisation 90% immunisation coverage No drop-out 80% Compliance Latrine coverage: 1 per family Global malnutrition rate less than 8%
DPT1 coverage = 82% DPT3 coverage = 70% DPT1 DPT3 = 12% Measles = 50% Fully immunised = 45% 1 per 20-25 people 10-15% of children have Zscore < -2 SD of reference mean
Good practices and coverage Mixed practices and coverage Mixed practices and coverage Poor practices and coverage Good practices and coverage Mixed practices and coverage
Graphs can also show the trends over time of the gaps between the actual practices and the set standards. Possible solutions can then be found to address the causes of gaps in service.
Programme input indicators

To assess availability of inputs consider the following: Availability of essential drugs; Daily staff attendance; Expenditures compared to budgeted amounts; Quality and use of food rations (food basket monitoring). To assess the adequacy of inputs, check whether: Enough stocks are maintained and that the use of supplies is monitored; Minimum standards are being followed, e.g. pesticides being sprayed are safe for the user and the environment; Local skills and resources (including traditional healers) are fully used (e.g. most of the critical jobs should be assigned to the affected population).
Scan all analysis results
Consistent-cross-check data gathered by different sources/methods to build a more accurate understanding of the results; Convincing-findings should be consistent with existing scientific knowledge of disaster experiences; Unbiased-search for systematic error at any stage of the study that produces results systematically differing from the true estimate due to selection, reporting or information bias. Do not generalise findings from a small area to the whole population: they might not be representative (e.g. hospital morbidity reports only represent those who use the services, not the entire population); Look for patterns among data variables: For example, if all children who were interviewed were weighed, check from the summary statistics and tables whether there are more ARI cases among underweight children compared to those with normal weight. Some of these patterns might become more obvious after graphing the data. Also consider the possibility of interaction between indicators (e.g. increased malaria incidence with season of the year, reduced water supply and under-five mortality, etc.).
Advanced data analysis

Sometimes, after carrying out special investigations or surveys, more advanced data analysis might be needed to determine the following:
Relative Risk (RR)

Relative risk compares the probability of disease in two different study populations: one with a risk factor and the other without. Relative risk is usually applied to longitudinal (cohort) studies. Relative risk can be calculated by determining the ratio of the proportion of exposed individuals with the event of interest out of the total exposed divided by the proportion of unexposed individuals with the event out of the total number of unexposed. The following table and example illustrate the concept:
Table 6-34 Calculation of relative risk and odds ratio
Status Diseased Non-diseased Total
Exposed Unexposed Total
A C a+c
B D b+d
a+b c+d T
Relative risk is: Relative risk = [a / (a + b)] / [c/(c + d)] For example, consider the following data obtained from a survey of incident malaria following a period of high malarial transmission. One purpose of the survey was to determine whether people who use Insecticide Treated bed Nets (ITN) have a lower risk for contracting malaria.
Before drawing conclusions about the data, scan all analysis results (summary statistics, tables, graphs, and indicators) as follows: Check that the results are:
278
Table 6-35 Calculation of relative risk for malaria among ITN Users versus non-ITN users
Status
Malaria +
Malaria -
Total
Use ITN Do Not Use ITN Total
20 40 60
30 10 40
50 50 100
Relative risk is: Relative risk = [20/(20 + 30)]/[40/(40 + 10)] = 0.5 This relative risk can be interpreted as follows. The risk of malaria among those who use ITNs is one-half the risk of malaria among those who do not use ITNs.
Odds Ratio (OR)

Odds ratio is used to compare the probability of having a risk factor in two different study populations: one with and the other without the disease. Odds ratio is usually calculated for case-control studies in a representative sample with results applied to the entire population. Odds Ratio can be calculated by determining the odds of a risk factor among individuals with the event of interest divided by the odds of a risk factor among individuals without the event of interest. Table above calculates odds ratio as a cross-product ratio: Odds ratio is: Odds ratio = (a d) / (b c) For example, consider the following data obtained from a community-based malaria prevalence survey. One purpose of the survey was to determine whether people with malaria are more likely to be under the age of five compared to those without malaria.
Table 6-36 Calculation of odds ratio for age less than five among those with and without malaria
Status Malaria + Malaria Total
Age < 5 Age 5 Total
40 20 60
10 30 40
50 50 100
Odds ratio is: Odds ratio = (40 30) / (20 10) = 6.0 This odds ratio can be interpreted as follows. The odds of being less than age of five years are six times greater among those with malaria compared to those without malaria. Alternatively, odds ratio can be interpreted in the following way. The odds of malaria are six times greater among those less than five years of age when compared to those aged five or older.
Sensitivity, specificity and predictive values

These tools assess whether a screening test or procedure can sufficiently detect a condition or if further tests are needed (e.g. clinical diagnosis of anaemia, pneumonia, etc.).
Confidence limits
The 95% confidence interval can be calculated using the following general formula: Point estimate [1.96 SE (estimate)] Where the point estimate is the relative risk or odds ratio, as calculated above, and SE (estimate) is the standard error of the relative risk or odds ratio. Referring to table 6-33, standard error for the relative risk can be calculated as follows, in the logarithmic scale: SE (log relative risk) = [b/(a(a+b))] + [d/(c(c+d))] Referring to table 6-33, standard error for the odds ratio can be calculated as follows, in the logarithmic scale: SE (log odds ratio) = (1/a + 1/b + 1/c + 1/d) Thus, the 95% Confidence Interval (CI) for odds ratio calculated above using table 4-33 is: 95% CI (log odds ratio) = log odds ratio [1.96 (1/40 + 1/10 + 1/20 + 1/30)] Employing basic algebraic techniques, we can calculate the 95% CI for odds ratio: 95% CI (odds ratio) = exp [log 6.0 [1.96 (1/40 + 1/10 + 1/20 + 1/30)]] Thus, the lower 95% confidence limit = 6.0 e-[1.96 (1/40 + 1/10 + 1/20 + 1/30)] = 2.44 Thus, the upper 95% confidence limit = 6.0 e [1.96 (1/40 + 1/10 + 1/20 + 1/30)] = 14.74 Using these confidence limits, we can say that the odds of malaria is six times greater among those less than five years compared to those greater than five with a 95% CI of 2.44 to 14.74. This means that true populations odds ratio would fall between 2.44 and 14.74 in 95% of all surveys conducted in a similar way among the same population. Although not shown here, similar calculations can determine the 95% confidence limits for Relative Risk using the standard error formula given above.
Significance tests
Significance tests (e.g., Chi2 test) are performed to establish whether two quite different factors, e.g. the diagnosis of anaemia and hookworm infection could be statistically associated, or whether the apparent relationship might have only occurred by chance. Results will be expressed as a statistical probability, where a P-value less than 0.05 implies association. Significance tests can also be used to establish whether the observed differences between different study populations are real or due to chance alone. For example, following a malaria survey, a resulting P-value of less than 0.05 would indicate that there is a real difference between the spleen rates of males and females that is not due to chance factors from sampling. However, this finding should be interpreted against existing scientific knowledge about malaria transmission before drawing the final conclusion of confirming the finding or repeating the survey in another population. The chi-squared statistic (X2) for a two-by-two contingency table can be calculated by hand as follows:
X2 = (ad bc) 2 (a+b+c+d) (a+b) (c+d) (b+d) (a+c)
with (r-1)(c-1) degrees of freedom.
Confidence limits indicate the probability (usually 95%) that the estimate obtained from the sample will not differ from the true population rate by more than the range defined by the confidence limit. For example, in a sample of 100 individuals in which a prevalence rate of 20% is observed, there is a 95% probability that the prevalence for the whole population will lie between (205) = 15% and (20 + 5) = 25%.
280
The following X2 can be calculated for table 6-33:
X2 = [(4030 1020)2 (40 + 10 + 20 + 30)]/[(40 + 10)(20 + 30)(10 + 30)(40 + 20)] = 16.667 The degrees of freedom (df) are (2 1) (2 1) = 1 Using a distribution table of chi-squared values (available in appendices in many statistical texts), the significance of this value can be determined. With a chi-squared value of 16.667 and 1 degrees of freedom (df), the obtained p-value would be less than or equal to 0.001. This means that the obtained Odds Ratio (6.0) is statistically significant. In other words, the higher malaria risk obtained among individuals less than five years of age compared to those aged five or more in this population is real and not due to chance. Note: Carrying out the above-described statistical analysis procedures in detail is beyond the scope of this book. If EPI-INFO is available, it can be used to perform these tests. Otherwise, please refer to standard statistical texts for full details.
Presenting data
Data can be presented in the form of tables or graphs because they create a clearer impression than numbers alone. However, basic rules should be followed when drawing and presenting graphs and tables, such as: Neatly draw and label all presentations and include a description of the data; Present only the most essential features in graphs. Otherwise, the simplicity and clarity of the information is destroyed; Limit the number of graphs, tables, etc. because too many can be confusing; Each presentation should be of reasonable size-not too big or too small; Use different colours/shadings/lines to increase contrast between data categories. This makes graphs easier to understand. A few tables were shown in the previous section on analysing and interpreting data. Graphs are very useful because they help define patterns in the data. The figures below show examples of graphs: Histogramto show the frequency distribution of large samples of quantitative data.
Incidence of ARI in Alpha camp by age and sex

45% 40% 35% 30% 25% 20% 15% 10% 5% 0% <5 >5
Males Females
Pie chartto show proportions of different segments of a whole, e.g. specific causes of death.
Time chartto show trends and changes in health of the population and disease occurrence over time. The time variable is usually placed on the x-axis and the frequency or rate on the y-axis.
Prevalence of diarrhoea among displaced population in Alpha camp 1995

80% 60% 40% 20% 0% Jan Feb Mar Apr May under five years over 5 years
Age-group
282
Scatter diagramto show relationship between a limited number of observations.
Average weight gain/day among under 5s enrolled in SFC at Alpha Camp
18
Ave weight gain (g/day

13
Females Total Males

0 2 4 6
Time (months)
Spot mapto show distribution of cases within an area.16
Communicating and using information

Figure 6-7: Flow of information in emergencies
International level
National level
Project level
Health facility
Community leaders, CHWs
Displaced population Information in emergencies must flow two ways. The objective of surveillance, surveys, outbreak investigations and health information is not simply to collect and report data. The objective is to improve how the relief programme is managed as well as provide feedback to data collectors so that they feel motivated. Programmes also have a responsibility to distribute information to affected populations themselves. After analysing data, the results should be communicated in a form that everyone can understand. The table below gives a summary of the reports and recipients of the information.
Table 6-37: Description of different reports and recipients
Type of report Recipient
Full detailed report for those in a position to improve situation and provide additional resources.
Decision makers at the national and international level so that appropriate control measures can be organised; Agencies and service providers of similar programmes; Senior health workers responsible for data collection to improve diagnosis and management of disease cases.
Summary of report for those Community health workers so that they will be who gave support or helped to motivated to continue collecting data. collect data. Very brief summary of the most important findings and conclusions. General population to be aware of health risks and to improve how they manage their illnesses at home.

Chapter 6 Epidemiology and Surveillance

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Chapter 6 Epidemiology and Surveillance

Diunggah oleh

Hak Cipta:

Format Tersedia

6

Epidemiology and surveillance

Public health guide for emergencies I 221

Basic principles of epidemiology in emergencies

what, who, where, when, how, why, and what next.

Epidemiology and surveillance

Epidemiology and surveillance

Epidemiology and surveillance

Role of epidemiology in emergencies

Public health guide for emergencies I 223

Constraints of epidemiology in emergencies

Key epidemiological indicators

Epidemiology and surveillance

Epidemiology and surveillance

Health policy (might be difficult to measure)

Public health guide for emergencies I 225

Estimating population size

Different methods for estimation

Acute emergency situations

Epidemiology and surveillance

Epidemiology and surveillance

Public health guide for emergencies I 227

International Federation refugee camp Photo: International Federation

Epidemiology and surveillance

Table 6-2: Camp register

Epidemiology and surveillance

Health problems Priority Addressed

Access to health facility

sanitation Immunisation Immunisation Malnutrition

Water source not reliable Easily floods in rainy season

Table 6-3: Community health worker household register

Household ID: 02/90/12 Head of household: Aladdin Ali

Sinbad Ali Baba

Poor access to formal health care, low literacy

Calculating rates, ratios and proportions

Rate = x/y x factor

Public health guide for emergencies I 229

Epidemiology and surveillance

Epidemiology and surveillance

10,000 # of days in time period

10,000 # of days in time period =

Public health guide for emergencies I 231

Crude mortality rate for population A

Crude mortality rate for Population B

Meningococcal meningitis is usually calculated per 100,000 persons/week

Epidemiology and surveillance

Epidemiology and surveillance

Public health guide for emergencies I 233

Number of days health facility open per week

Epidemiology and surveillance

Epidemiology and surveillance

Interpretation of epidemiological data

Public health guide for emergencies I 235

Epidemiology and surveillance

Rapid health assessment

Epidemiology and surveillance

Objectives of rapid health needs assessment

The magnitude of the emergency and population size affected;

Preparing for rapid health needs assessment

Public health guide for emergencies I 237

Epidemiology and surveillance

Epidemiology and surveillance

Defining the information needed